Analysis of Esac-Net/Ears-Net Data from 29 Eea Countries for Associations Between Patterns of Antimicrobial Consumption and Resistance – Implications for Antimicrobial Stewardship?

1. Introduction

It is forecast that antimicrobial resistance will account for 10 million deaths per annum by 2050 unless countered by radical action [1]. Consensus dictates that improvident use of antimicrobials is liable to hasten selection of resistant pathogens, but it is recognised that some agents have greater ecological impact than others [1,2,3,4,5,6]. Drugs with a proclivity to quickly select for resistance after even limited use, can be seen as having a high resistance potential [6,7]. Conversely, agents which select for resistance only after heavy use can be considered as having a low resistance potential [6,7]. Emphatically, this concept is a gross generalisation with critical exceptions; no drug thus far has been ‘resistance-proof’, and resistance will inevitably emerge if consumption is sustained beyond a certain threshold [8,9,10,11]. Lavish use of any antibiotic, no matter how low its perceived resistance potential, should therefore be discouraged. Pathogens vary in their capacity to acquire resistance to specific agents [6,7,8,9,10]. We are only now, ca.80 years after the clinical debut of penicillin, seeing the first signs of resistance developing in Streptococcus pyogenes [12,13]. A distinct scenario occurred with Staphylococcus aureus, penicillin resistant strains of which emerged and spread swiftly, gaining dominance first in hospitals and then the wider community [14,15,16]. Regarding the rapidity with which penicillin resistance was acquired, most other inherently sensitive pathogens including pneumococci, gonococci and meningococci fell between these superlatives but, overall, resistance emerged only gradually after decades of intense use [17,18,19]. Penicillin could therefore be viewed as possessing a low resistance potential. Contrastingly, 3rd generation cephalosporins were quickly met with resistance problems in the form of nosocomial outbreaks due to multi-resistant Gram-negative bacilli (MDR-GNB) [20,21,22,23,24,25,26]. These agents also became associated with methicillin resistant S. aureus (MRSA) and vancomycin resistant Enterococci (VRE), more so than did the anti-staphylococcal penicillins and vancomycin themselves, respectively [26,27,28,29,30,31,32,33]. Resistance to the prototype fluoroquinolone, ciprofloxacin, by epidemic MRSA (EMRSA-15) clone ST22-A2, arose rapidly after its introduction in the UK and was a pivotal factor in its successful pandemic spread [34]. Moreover, 3rd generation cephalosporins and fluoroquinolones each pose distinctly high risks of promoting Clostridioides difficile infection [26,35,36,37,38,39]. Considering the aforementioned factors, these drugs could be construed as exhibiting a high resistance potential. The question of which, if any, antibiotics should be categorised as having high or low potential for selecting resistance has been much debated [5,6,7,8,40]. Generally, narrow spectrum antibiotics are lower risk whilst broader spectrum agents are higher risk though there may be exceptions [2,3,5,40]. The term ‘broad spectrum’ can be ambiguous [41,42]. Breadth of spectrum has historically been defined in terms of Gram-stain and clinical versatility whereby the spectrum of activity incorporates any organisms against which a given agent is active at clinically achievable concentrations [41,42]. It does not automatically follow, however, that the extent to which a given antimicrobial selects problematic resistance or perturbs the microbiome is proportional to its breadth of spectrum as judged by this metric [6,7,40,41,42]. As a case in point, doxycycline has been advocated in recent antimicrobial guidelines as a drug with comparatively limited adverse ecological impact, and a minimal risk of promoting C. difficile colitis [43,44,45,46,47,48,49]. Whilst doxycycline poses less risk than some other agents, it, like other tetracyclines, is assuredly not ‘narrow spectrum’ in the conventional sense as outlined here [49]. Indeed, the term ‘broad spectrum’ was first coined in the 1940s to describe the first tetracyclines and chloramphenicol, given their expansive utility to treat infections due to both Gram-positive and Gram-negative pathogens [49,50,51,52]. Another example of an antibiotic with comparatively low resistance potential despite possessing broad activity is piperacillin-tazobactam. This drug has a wider spectrum of antibacterial action than 3rd generation cephalosporins yet, relatively speaking, demonstrates lower propensity for selecting C. difficile, VRE and Enterobacterales harbouring extended spectrum β-lactamases (ESBLs) and/or derepressed AmpC β-lactamases [53,54,55,56,57,58,59]. The World Health Organization (WHO) has adapted the concept of resistance potential in the form of the Access, Watch, Reserve (AWaRe) schema which stratifies antimicrobials by risk [1,2,3,4,5]. Access antibiotics are those recommended by the WHO for routine management of infections [1,2,3,4,5]. They are considered to have a low resistance potential. Watch agents are generally broader spectrum with higher resistance potential and recommended only when access agents are unsuitable, for instance, because of allergy or resistance. Reserve agents are used only for multidrug resistant infections, where use of drugs from the former two categories is precluded [1,2,3,4,5]. Reserve agents are often expensive, newer and often have high toxicity and/or resistance potential [1,2,3,4,5]. Drugs from the same chemical class might have divergent AWaRe classifications [1,2,3,4,5]. The legitimacy of individual AWaRe groupings has been questioned [40]. Most macrolides, for instance, are allocated to the watch group despite having an arguably narrower spectrum of activity than some access drugs, such as amoxicillin-clavulanate [40,60]. Likewise, some argue against the inclusion of lincosamides and nitroimidazoles in the access group as their activity against anaerobes has been linked with microbiome disruption and reduced resilience to colonisation by resistant organisms including VRE and multidrug resistant Enterobacterales [40,61,62,63,64,65,66,67,68,69]. Furthermore, lincosamides also have the strongest association with C. difficile colitis of any antibiotic class deployed clinically (70,71,72). The European Antimicrobial Resistance Surveillance Network (EARS-NET) and European Surveillance of Antimicrobial Consumption Network (ESAC-NET) curate population level data over time and throughout the European Economic Area (EEA). EARS-NET monitors resistance to antimicrobials amongst invasive isolates of sentinel pathogens whereas ESAC-NET tracks usage of systemic antimicrobials. Together, data from these networks can be interrogated to determine whether spatiotemporal correlations between antimicrobial consumption and resistance exist. Using univariate and multivariate regression analyses, this was assessed methodically for 29 EEA countries.

2. Methods

Antimicrobial consumption data in the form of defined daily doses per 1000 inhabitants per day (ddd/1000/day) were collated from ESAC-NET for 29 countries using 18 Anatomical Therapeutic Chemical Classification (ATCC) Codes. ATCC codes corresponded to antibiotic classes as follows: J01 total systemic antibacterials, J01A tetracyclines, J01CA extended spectrum penicillins, J01CE β-lactamase labile penicillins, J01CF β-lactamase stable penicillins, J01CR penicillin / β-lactamase inhibitor combinations, J01DB first generation cephalosporins, J01DC second generation cephalosporins, J01DD third generation cephalosporins, J01 DH carbapenems, J01E sulphonamides & trimethoprim, J01FA macrolides, J01FF lincosamides, J01G aminoglycosides, J01M quinolones, J01XA glycopeptides, J01XD nitroimidazoles and J01XE nitrofurans. Note that for tetracyclines (J01A), sulphonamides & trimethoprim (J01E) and quinolones (J01M) consumption was resolved only to ATCC level 3 whilst other classes were subdivided down to ATCC level 4. Data were not available at ATCC level 5, corresponding to individual compounds. The 29 nations included were as follows: Austria, Belgium, Bulgaria, Croatia, Cyprus, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden and United Kingdom. Mean use as ddd/1000/day/ATCC code was quantified for each of these 29 countries over 4 periods: 2017-2018, 2018-2019, 2019-2020 and 2020-2021. Mean values over two-year periods were used to account for temporal lags in resistance behind fluctuations in consumption. Percentages of invasive isolates in each country over years 2018, 2019, 2020 and 2021 were recorded for resistotypes relevant to each of 4 pathogens: Escherichia coli (E. coli), Klebsiella pneumoniae (K. pneumoniae), MRSA and VRE. Resistotypes considered were as follows: - For E. coli: aminopenicillin (AMPR), 3rd generation cephalosporin (3GCR), fluoroquinolone (FQR), aminoglycoside (AGR), triple resistance to 3rd generation cephalosporin, fluoroquinolone and aminoglycoside (3XR). For K. pneumoniae: carbapenem (CARBR), 3GCR, FQR, AGR, 3XR. For S. aureus: methicillin resistance (MRSA). For Enterococcus faecium: vancomycin resistance (VRE). Resistance data for the first and last years, 2018 and 2021, were presented graphically. Consumption data for Austria were only available for the 2019-2020 and 2021-2022 periods. Consumption data for Czechia was only available for 2021-2022 and UK consumption data were available for 2017-2018 and 2018-2019 only. Data for AMPR E. coli was not available from Sweden. Univariate and multivariate regressions were modelled in Microsoft Excel spreadsheets for each resistotype in each pathogen versus usage in ddd/1000/day for each country. Backwards stepwise selection was used to select variables which best fit final multivariate regression models for each species and resistotype spanning the whole 5-year time series and 29 nations. Antimicrobial classes represented by each ATCC code were ranked by their correlation R, from lowest to highest risk for each resistotype in each pathogen and presented in tabulated format, for univariate and multivariate analyses. Sample sizes were calculated a priori to ensure power of each multivariate model was ≥ 0.8. Variance inflation factors (VIF) were calculated for each multivariate model to rule out multicollinearity. Consumption in the 2017/2018 and 2020/2021 periods was plotted graphically for each of 7 countries with decreases in ≥ 2 resistotypes and no increases in any of 12 resistotypes (group 1 countries) alongside consumption over the same period for each of 7 countries with increases in ≥ 2 resistotypes (group 2 countries).

3. Results

Antimicrobial use trends for each EEA nation over the 5 years analysed are presented numerically as ddd/1000/day (Table 1). Resistance levels in each EEA country are plotted graphically for the first (2018) and last (2021) of the 4 years examined for E. coli (Figure 1), K. pneumoniae (Figure 2), VRE and MRSA (Figure 3). Consumption of each antimicrobial class in the 2017/2018 and 2020/2021 seasons by group 1 and group 2 countries is also described in the following text and plotted in Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8. Associations between each resistotype in each pathogen with volume of consumption for each antimicrobial class are tabulated and heat mapped by order of strength in univariate analyses (Table 2). The subset of associations which retained statistical significance in multivariate analyses are similarly tabulated (Table 3) and discussed in the following paragraphs.

Table 1. Consumption in ddd/1000/day for each antimicrobial class in EEA countries.

J01	17/18	18/19	19/20	20/21	J01A	17/18	18/19	19/20	20/21	J01CA	17/18	18/19	19/20	20/21
AT	NA	NA	10.194	8.809	AT	NA	NA	0.353	0.350	AT	NA	NA	0.789	0.718
BE	22.531	21.833	19.012	17.025	BE	1.911	1.884	1.761	1.689	BE	4.838	4.883	4.047	3.284
BG	20.800	20.910	21.707	22.077	BG	1.639	1.665	1.964	2.326	BG	2.901	2.791	2.712	2.704
HR	18.685	18.793	17.221	16.903	HR	1.022	0.988	0.892	0.866	HR	1.914	1.803	1.484	1.224
CY	28.436	29.055	29.522	26.979	CY	3.657	3.797	3.565	3.481	CY	1.923	1.663	1.385	1.239
CZ	NA	NA	NA	13.540	CZ	NA	NA	NA	1.754	CZ	NA	NA	NA	0.544
DK	15.899	15.427	14.780	14.335	DK	1.454	1.491	1.638	1.727	DK	3.688	3.646	3.563	3.513
EE	11.682	11.774	11.117	10.263	EE	1.299	1.331	1.387	1.375	EE	1.648	1.667	1.503	1.314
FI	15.573	15.050	13.273	11.588	FI	3.330	3.185	2.826	2.447	FI	2.691	2.667	2.238	1.815
FR	25.022	25.216	22.710	20.938	FR	3.102	3.208	3.102	3.010	FR	8.622	8.929	7.712	6.566
DE	21.674	21.251	19.328	17.911	DE	2.345	2.335	2.026	1.739	DE	4.119	4.029	3.451	2.988
GR	34.176	34.110	31.100	25.829	GR	2.924	3.049	3.070	3.037	GR	4.619	4.619	4.275	3.800
HU	14.699	14.632	12.835	11.594	HU	1.121	1.188	1.229	1.241	HU	0.750	0.680	0.535	0.411
IS	20.546	19.835	17.889	16.618	IS	5.221	5.285	5.094	4.740	IS	3.946	4.128	3.729	3.473
IE	21.694	22.618	20.684	18.181	IE	2.747	2.847	3.028	3.018	IE	3.375	3.758	3.481	2.982
IT	21.156	21.545	20.051	17.972	IT	0.529	0.571	0.608	0.639	IT	2.061	2.033	1.785	1.507
LV	13.860	13.842	12.898	11.768	LV	2.252	2.280	2.383	2.514	LV	3.043	2.822	2.341	1.874
LT	16.240	16.105	15.138	13.933	LT	1.421	1.431	1.506	1.504	LT	5.119	4.961	4.374	3.742
LU	22.316	21.608	18.621	15.999	LU	1.594	2.070	1.830	1.601	LU	3.100	3.439	2.974	2.475
MT	21.396	20.461	18.628	16.170	MT	1.694	1.736	2.047	2.224	MT	0.619	0.503	0.451	0.423
NL	9.755	9.605	9.004	8.431	NL	1.983	1.905	1.706	1.530	NL	1.372	1.358	1.175	1.030
NO	15.510	15.095	14.415	13.953	NO	2.748	2.739	2.666	2.553	NO	2.151	2.133	2.040	1.909
PL	24.895	24.015	21.077	19.358	PL	2.395	2.273	2.008	1.790	PL	3.522	3.346	2.677	2.170
PT	18.684	19.182	17.231	15.222	PT	0.857	0.875	0.854	0.862	PT	1.909	2.001	1.646	1.286
RO	24.801	25.432	25.471	25.419	RO	0.898	0.931	0.983	1.053	RO	3.501	3.298	3.041	2.802
SK	20.985	20.671	16.884	15.194	SK	1.702	1.735	1.711	1.722	SK	0.782	0.633	0.466	0.355
SI	13.132	13.080	11.581	10.165	SI	0.512	0.547	0.553	0.548	SI	2.339	2.319	1.848	1.420
ES	26.476	25.547	22.322	19.878	ES	1.488	1.502	1.499	1.534	ES	6.059	5.913	4.990	4.139
SE	12.603	12.117	11.068	10.198	SE	2.422	2.304	2.170	2.027	SE	1.210	1.209	1.157	1.092
UK	19.061	18.470	NA	NA	UK	4.970	4.937	NA	NA	UK	3.462	3.331	NA	NA
J01CE	17/18	18/19	19/20	20/21	J01CF	17/18	18/19	19/20	20/21	J01CR	17/18	18/19	19/20	20/21
AT	NA	NA	0.47865	0.31515	AT	NA	NA	0.029	0.028	AT	NA	NA	3.340	2.949
BE	0.04	0.0396	0.0284	0.0204	BE	0.359	0.372	0.368	0.364	BE	5.433	5.400	4.628	4.075
BG	0.13	0.12625	0.07255	0.01505	BG	0.000	0.000	0.000	0.000	BG	2.530	2.646	2.373	2.130
HR	0.51	0.4852	0.36315	0.24515	HR	0.043	0.048	0.046	0.042	HR	5.822	6.011	5.577	5.237
CY	0.36	0.05745	0.1339	0.1754	CY	0.025	0.028	0.025	0.021	CY	7.095	7.557	7.475	4.643
CZ	NA	NA	NA	0.94935	CZ	NA	NA	NA	0.08595	CZ	NA	NA	NA	2.943
DK	3.89	3.66995	3.2711	2.9866	DK	1.775	1.794	1.753	1.753	DK	1.018	0.966	0.903	0.861
EE	0.19	0.17135	0.141	0.1145	EE	0.123	0.113	0.098	0.094	EE	2.173	2.332	2.281	2.230
FI	1.22	1.22375	1.08875	0.94825	FI	0.161	0.185	0.194	0.201	FI	0.872	0.865	0.751	0.623
FR	0.18	0.1848	0.184	0.1775	FR	0.176	0.163	0.144	0.129	FR	5.028	4.900	4.454	4.173
DE	0.51	0.50125	0.4148	0.3213	DE	0.543	0.523	0.372	0.242	DE	4.507	4.486	4.269	4.120
GR	0.31	0.0327	0.03035	0.04235	GR	0.014	0.016	0.016	0.016	GR	6.513	6.648	5.802	4.830
HU	0.13	0.1183	0.0851	0.0516	HU	0.000	0.000	0.001	0.001	HU	3.972	4.095	3.518	2.957
IS	1.87	1.74915	1.56295	1.37995	IS	1.164	1.158	1.081	1.063	IS	2.783	2.380	1.885	1.647
IE	1.20	1.2331	1.08235	1.27885	IE	1.534	1.542	1.521	1.518	IE	4.323	4.203	3.389	2.648
IT	0.00	0.0011	0.0008	0.0003	IT	0.023	0.019	0.012	0.014	IT	7.334	7.590	7.024	6.247
LV	0.09	0.07285	0.0599	0.0469	LV	0.048	0.053	0.055	0.054	LV	1.862	2.092	2.077	1.974
LT	0.29	0.28055	0.21975	0.13475	LT	0.005	0.000	0.000	0.000	LT	1.877	1.823	1.649	1.801
LU	0.02	0.0127	0.0106	0.0136	LU	0.326	0.202	0.187	0.175	LU	4.746	5.257	4.469	3.808
MT	0.21	0.1853	0.211	0.15	MT	0.364	0.196	0.175	0.166	MT	6.606	6.343	5.938	5.334
NL	0.17	0.144	0.1676	0.1476	NL	0.580	0.588	0.580	0.578	NL	1.077	1.056	0.971	0.905
NO	3.17	3.1575	2.8395	2.473	NO	0.795	0.832	0.856	0.867	NO	0.073	0.082	0.091	0.100
PL	0.34	0.3534	0.26615	0.22665	PL	0.023	0.031	0.035	0.037	PL	3.163	3.151	2.692	2.418
PT	0.03	0.0278	0.02495	0.02245	PT	0.848	0.659	0.400	0.355	PT	6.637	6.952	6.307	5.513
RO	0.60	0.57645	0.5444	0.4154	RO	0.461	0.440	0.401	0.358	RO	6.733	7.179	7.150	7.116
SK	1.07	0.9965	0.74005	0.54005	SK	0.000	0.000	0.000	0.000	SK	3.505	3.613	3.142	2.655
SI	1.61	1.532	1.1214	0.7219	SI	0.286	0.320	0.323	0.306	SI	3.276	3.262	3.031	2.814
ES	0.11	0.11395	0.0944	0.0639	ES	0.272	0.267	0.241	0.217	ES	8.179	7.770	6.732	5.968
SE	3.34	3.14025	2.57685	2.16935	SE	1.800	1.793	1.723	1.662	SE	0.318	0.337	0.337	0.328
UK	0.93	0.8796	NA	NA	UK	1.882	1.846	NA	NA	UK	1.118	1.114	NA	NA
J01DB	17/18	18/19	19/20	20/21	J01DC	17/18	18/19	19/20	20/21	J01DD	17/18	18/19	19/20	20/21
AT	NA	NA	0.40155	0.37505	AT	NA	NA	0.886	0.715	AT	NA	NA	0.261	0.259
BE	0.19	0.1789	0.15235	0.14985	BE	1.233	1.256	1.042	0.837	BE	0.098	0.101	0.098	0.093
BG	0.14	0.1215	0.13845	0.16245	BG	3.432	3.473	3.035	2.691	BG	1.597	1.661	1.836	2.373
HR	0.52	0.4591	0.39105	0.35905	HR	1.904	1.927	1.728	1.577	HR	0.519	0.604	0.635	0.727
CY	0.22	0.1675	0.12565	0.10365	CY	4.445	4.700	4.868	4.286	CY	1.048	1.131	1.153	1.112
CZ	NA	NA	NA	0.0925	CZ	NA	NA	NA	1.5863	CZ	NA	NA	NA	0.181
DK	0.004	0.00345	0.0027	0.0027	DK	0.215	0.184	0.164	0.152	DK	0.030	0.029	0.028	0.026
EE	0.246	0.2467	0.2302	0.2047	EE	1.342	1.317	1.243	1.109	EE	0.054	0.053	0.056	0.058
FI	2.016	1.9312	1.7145	1.5445	FI	0.683	0.712	0.616	0.569	FI	0.086	0.076	0.083	0.080
FR	0.105	0.1048	0.100	0.100	FR	0.328	0.285	0.210	0.144	FR	1.340	1.226	1.041	0.913
DE	0.188	0.17525	0.15575	0.14375	DE	1.276	1.266	1.211	1.086	DE	0.785	0.793	0.771	0.780
GR	0.006	0.00505	0.0032	0.0022	GR	7.781	7.673	6.586	4.950	GR	0.329	0.398	0.418	0.447
HU	0.040	0.0429	0.04185	0.03935	HU	1.902	1.858	1.519	1.157	HU	0.486	0.518	0.502	0.494
IS	0.692	0.6533	0.60755	0.61255	IS	0.031	0.036	0.029	0.016	IS	0.129	0.121	0.121	0.128
IE	0.409	0.4487	0.47295	0.54895	IE	0.738	0.744	0.635	0.455	IE	0.100	0.101	0.091	0.086
IT	0.136	0.1399	0.1329	0.1249	IT	0.205	0.209	0.177	0.132	IT	2.026	2.208	2.056	1.743
LV	0.288	0.25935	0.2358	0.1998	LV	0.493	0.509	0.435	0.353	LV	0.450	0.447	0.469	0.442
LT	0.464	0.4213	0.3509	0.2624	LT	1.414	1.518	1.544	1.517	LT	0.170	0.160	0.143	0.126
LU	0.106	0.09325	0.08035	0.07135	LU	2.554	2.905	2.351	1.776	LU	0.350	0.179	0.183	0.181
MT	0.054	0.06425	0.0612	0.0557	MT	2.270	2.182	1.880	1.317	MT	0.689	0.693	0.659	0.491
NL	0.072	0.07305	0.0668	0.0648	NL	0.086	0.090	0.089	0.081	NL	0.074	0.078	0.085	0.096
NO	0.145	0.142	0.1375	0.137	NO	0.021	0.019	0.017	0.011	NO	0.133	0.129	0.116	0.113
PL	0.237	0.145	0.09195	0.07345	PL	3.580	3.359	3.070	2.350	PL	0.200	0.226	0.263	0.364
PT	0.289	0.24885	0.2074	0.1784	PT	1.234	1.315	1.286	1.238	PT	0.415	0.377	0.327	0.339
RO	0.203	0.1734	0.1289	0.1024	RO	3.485	3.615	3.336	2.939	RO	1.319	1.426	1.395	1.430
SK	0.256	0.249	0.19975	0.15125	SK	3.937	3.998	3.371	2.776	SK	1.616	1.689	0.860	0.863
SI	0.136	0.1423	0.12885	0.11085	SI	0.301	0.301	0.366	0.431	SI	0.145	0.155	0.137	0.120
ES	0.145	0.1473	0.14245	0.14395	ES	1.763	1.776	1.590	1.423	ES	0.698	0.726	0.704	0.676
SE	0.077	0.06735	0.0583	0.0543	SE	0.010	0.010	0.010	0.010	SE	0.113	0.112	0.107	0.106
UK	0.216	0.20195	NA	NA	UK	0.043	0.042	NA	NA	UK	0.069	0.073	NA	NA
J01DH	17/18	18/19	19/20	20/21	J01E	17/18	18/19	19/20	20/21	J01FA	17/18	18/19	19/20	20/21
AT	NA	NA	0.10945	0.10945	AT	NA	NA	0.267	0.270	AT	NA	NA	1.293	0.950
BE	0.04925	0.0482	0.046	0.0445	BE	0.236	0.254	0.262	0.269	BE	3.167	3.196	2.755	2.349
BG	0.0295	0.036	0.06985	0.11035	BG	0.862	0.826	0.715	0.650	BG	3.232	3.299	4.124	4.906
HR	0.07515	0.07845	0.08535	0.10285	HR	0.562	0.532	0.499	0.480	HR	2.535	2.546	2.355	2.478
CY	0.13075	0.14005	0.14505	0.16955	CY	0.276	0.228	0.239	0.266	CY	2.851	2.968	2.777	2.573
CZ	NA	NA	NA	0.08925	CZ	NA	NA	NA	0.7719	CZ	NA	NA	NA	2.380
DK	0.0546	0.05395	0.0536	0.0521	DK	0.806	0.750	0.692	0.669	DK	1.608	1.520	1.370	1.210
EE	0.05635	0.06235	0.0726	0.0796	EE	0.474	0.469	0.396	0.346	EE	2.279	2.313	2.058	1.749
FI	0.05465	0.04695	0.0414	0.0424	FI	1.084	1.111	1.095	1.029	FI	0.577	0.523	0.405	0.293
FR	0.03375	0.03325	0.032	0.0345	FR	0.472	0.470	0.494	0.520	FR	1.943	1.832	1.662	1.563
DE	0.0501	0.0525	0.0569	0.0639	DE	0.659	0.657	0.632	0.596	DE	2.951	2.881	2.571	2.366
GR	0.16865	0.15855	0.16435	0.19385	GR	0.329	0.332	0.403	0.431	GR	6.391	6.180	5.584	4.133
HU	0.04705	0.05015	0.05815	0.06665	HU	0.474	0.474	0.444	0.410	HU	2.339	2.338	2.094	2.183
IS	0.0242	0.03295	0.0315	0.0345	IS	0.551	0.570	0.554	0.542	IS	1.601	1.479	1.255	1.078
IE	0.04635	0.04625	0.0463	0.0468	IE	0.993	1.076	1.090	1.089	IE	4.247	4.079	3.498	2.787
IT	0.0381	0.0474	0.05435	0.05885	IT	0.816	0.934	1.033	1.046	IT	3.962	4.122	4.036	3.615
LV	0.03135	0.03235	0.0426	0.0461	LV	0.863	0.836	0.700	0.533	LV	2.052	2.128	1.896	1.630
LT	0.03935	0.0417	0.04655	0.05405	LT	0.688	0.770	0.748	0.554	LT	2.139	2.159	1.956	1.639
LU	0.05725	0.05285	0.04985	0.04985	LU	0.451	0.319	0.339	0.347	LU	3.714	2.987	2.533	2.099
MT	0.15435	0.08845	0.08465	0.08765	MT	0.410	0.407	0.399	0.386	MT	4.518	4.337	3.365	2.380
NL	0.01505	0.01485	0.0148	0.0148	NL	0.454	0.470	0.480	0.478	NL	1.224	1.247	1.200	1.122
NO	0.02625	0.0247	0.02395	0.02445	NO	0.728	0.764	0.775	0.765	NO	0.791	0.723	0.610	0.468
PL	0.0244	0.0284	0.03365	0.03965	PL	0.552	0.575	0.514	0.429	PL	4.507	4.250	2.687	2.571
PT	0.08185	0.07795	0.0781	0.0856	PT	0.446	0.424	0.428	0.429	PT	2.645	2.985	2.667	2.053
RO	0.0572	0.06745	0.08095	0.10095	RO	0.855	0.843	0.798	0.742	RO	2.773	2.883	3.721	4.515
SK	0.0425	0.0449	0.04495	0.05245	SK	0.406	0.445	0.462	0.451	SK	4.286	4.222	3.535	3.426
SI	0.05845	0.05415	0.053	0.058	SI	0.741	0.713	0.630	0.583	SI	1.605	1.628	1.412	1.182
ES	0.0877	0.0866	0.08735	0.09385	ES	0.467	0.465	0.454	0.458	ES	2.976	2.848	2.386	1.997
SE	0.0394	0.0403	0.04045	0.04145	SE	0.313	0.322	0.320	0.312	SE	0.239	0.243	0.271	0.265
UK	0.0524	0.0504	NA	NA	UK	1.069	0.942	NA	NA	UK	2.977	2.772	NA	NA
J01FF	17/18	18/19	19/20	20/21	J01G	17/18	18/19	19/20	20/21	J01M	17/18	18/19	19/20	20/21
AT	NA	NA	0.62585	0.58085	AT	NA	NA	0.028	0.026	AT	NA	NA	0.799	0.683
BE	0.4319	0.4505	0.43235	0.40235	BE	0.034	0.034	0.033	0.031	BE	1.847	1.031	0.662	0.590
BG	0.79515	0.82685	0.913	0.8825	BG	0.254	0.256	0.256	0.255	BG	2.996	2.963	3.303	3.820
HR	0.41405	0.43615	0.44415	0.46515	HR	0.098	0.095	0.081	0.075	HR	1.727	1.662	1.509	1.482
CY	0.11645	0.1413	0.122	0.1225	CY	0.078	0.079	0.084	0.091	CY	5.751	5.558	6.580	6.679
CZ	NA	NA	NA	0.40275	CZ	NA	NA	NA	0.081	CZ	NA	NA	NA	0.596
DK	0.06945	0.0714	0.07505	0.08005	DK	0.056	0.057	0.053	0.047	DK	0.565	0.519	0.465	0.433
EE	0.177	0.1761	0.1723	0.1668	EE	0.037	0.028	0.024	0.024	EE	0.924	0.817	0.828	0.838
FI	0.27765	0.27225	0.24955	0.23155	FI	0.015	0.017	0.016	0.015	FI	0.854	0.720	0.584	0.533
FR	0.1276	0.14795	0.1668	0.1768	FR	0.076	0.070	0.064	0.059	FR	1.543	1.452	1.311	1.225
DE	0.24335	0.25425	0.2683	0.2778	DE	0.090	0.086	0.076	0.068	DE	1.890	1.710	1.536	1.501
GR	0.4158	0.44065	0.41455	0.34855	GR	0.139	0.142	0.134	0.126	GR	2.979	3.197	3.019	2.691
HU	0.5974	0.596	0.5788	0.5608	HU	0.032	0.028	0.033	0.033	HU	2.554	2.285	1.808	1.544
IS	0.21445	0.19975	0.17445	0.16445	IS	0.026	0.026	0.028	0.029	IS	0.909	0.761	0.574	0.523
IE	0.1035	0.1024	0.1003	0.0973	IE	0.101	0.103	0.091	0.077	IE	0.884	0.745	0.552	0.465
IT	0.0312	0.03375	0.0339	0.0314	IT	0.068	0.063	0.057	0.056	IT	3.043	2.628	2.059	1.848
LV	0.14955	0.15445	0.15755	0.15555	LV	0.076	0.069	0.063	0.056	LV	1.249	1.148	0.998	0.898
LT	0.02825	0.02675	0.02495	0.02345	LT	0.069	0.065	0.057	0.049	LT	1.086	1.042	0.972	0.885
LU	0.89745	0.5109	0.47785	0.45385	LU	0.053	0.044	0.032	0.022	LU	2.585	1.963	1.551	1.384
MT	0.18435	0.21595	0.2191	0.2351	MT	0.081	0.089	0.076	0.057	MT	2.540	2.360	1.879	1.595
NL	0.246	0.2546	0.2562	0.2592	NL	0.055	0.052	0.047	0.043	NL	0.816	0.777	0.723	0.703
NO	0.22275	0.2192	0.20965	0.19765	NO	0.087	0.093	0.090	0.084	NO	0.380	0.338	0.293	0.265
PL	0.8168	0.80095	0.74665	0.73715	PL	0.078	0.074	0.065	0.058	PL	1.660	1.588	1.406	1.328
PT	0.0739	0.0771	0.0759	0.0809	PT	0.060	0.058	0.059	0.060	PT	1.839	1.728	1.466	1.286
RO	0.17485	0.19395	0.20495	0.20345	RO	0.204	0.191	0.172	0.154	RO	3.274	3.310	3.223	3.228
SK	0.5469	0.56615	0.5575	0.5365	SK	0.103	0.102	0.077	0.072	SK	2.371	2.003	1.490	1.327
SI	0.2569	0.25855	0.23775	0.21875	SI	0.061	0.061	0.058	0.054	SI	1.305	1.234	1.069	0.973
ES	0.21775	0.21625	0.1997	0.2062	ES	0.081	0.079	0.077	0.078	ES	3.042	2.746	2.250	1.955
SE	0.34235	0.32515	0.29815	0.27315	SE	0.026	0.024	0.023	0.022	SE	0.762	0.721	0.659	0.623
UK	0.100	0.100	NA	NA	UK	0.129	0.132	NA	NA	UK	0.568	0.542	NA	NA
J01XA	17/18	18/19	19/20	20/21	J01XD	17/18	18/19	19/20	20/21	J01XE	17/18	18/19	19/20	20/21
AT	NA	NA	0.02215	0.02365	AT	NA	NA	0.056	0.0552	AT	NA	NA	0.272	0.229
BE	0.042	0.0415	0.0407	0.0422	BE	0.020	0.020	0.019	0.018	BE	2.373	2.403	2.398	2.354
BG	0.02415	0.02355	0.017	0.021	BG	0.136	0.140	0.114	0.080	BG	0.000	0.000	0.000	0.000
HR	0.04565	0.04855	0.0482	0.0577	HR	0.129	0.133	0.110	0.113	HR	0.735	0.813	0.845	0.874
CY	0.085	0.09665	0.1032	0.1032	CY	0.196	0.242	0.254	0.191	CY	0.443	0.460	0.439	0.408
CZ	NA	NA	NA	0.0321	CZ	NA	NA	NA	0.067	CZ	NA	NA	NA	0.912
DK	0.02485	0.0246	0.0248	0.0248	DK	0.087	0.081	0.075	0.070	DK	0.215	0.217	0.274	0.278
EE	0.02235	0.02385	0.02385	0.02635	EE	0.044	0.042	0.040	0.035	EE	0.573	0.582	0.526	0.495
FI	0.0313	0.02735	0.0239	0.0269	FI	0.032	0.027	0.029	0.049	FI	0.405	0.310	0.291	0.314
FR	0.0371	0.0355	0.03315	0.03315	FR	0.054	0.052	0.050	0.051	FR	0.209	0.177	0.140	0.117
DE	0.05065	0.04965	0.0434	0.0404	DE	0.053	0.055	0.051	0.048	DE	0.811	0.824	0.845	0.805
GR	0.14485	0.1488	0.1518	0.1633	GR	0.187	0.169	0.147	0.124	GR	0.656	0.601	0.550	0.566
HU	0.0228	0.0216	0.02195	0.02445	HU	0.043	0.045	0.059	0.071	HU	0.085	0.183	0.193	0.191
IS	0.0322	0.03195	0.0171	0.0021	IS	0.054	0.052	0.051	0.052	IS	0.682	0.515	0.492	0.497
IE	0.0754	0.0773	0.074	0.068	IE	0.056	0.057	0.050	0.044	IE	0.691	1.369	1.383	1.376
IT	0.0602	0.05415	0.04905	0.04505	IT	0.031	0.045	0.047	0.041	IT	0.191	0.205	0.222	0.236
LV	0.0195	0.0309	0.04045	0.03295	LV	0.144	0.134	0.130	0.120	LV	0.711	0.731	0.769	0.807
LT	0.035	0.03895	0.04645	0.04745	LT	0.089	0.101	0.097	0.057	LT	1.222	1.279	1.403	1.439
LU	0.0382	0.03255	0.0315	0.0325	LU	0.037	0.038	0.036	0.033	LU	0.885	1.293	1.281	1.268
MT	0.12625	0.13565	0.14965	0.16315	MT	0.109	0.112	0.098	0.083	MT	0.428	0.490	0.697	0.811
NL	0.01995	0.01985	0.02025	0.02075	NL	0.034	0.034	0.033	0.032	NL	1.371	1.343	1.287	1.257
NO	0.01495	0.01495	0.01475	0.01425	NO	0.039	0.039	0.038	0.035	NO	0.244	0.214	0.218	0.223
PL	0.01825	0.0198	0.02265	0.03615	PL	0.072	0.078	0.079	0.083	PL	3.634	3.633	4.336	4.349
PT	0.04605	0.03855	0.0382	0.0372	PT	0.025	0.000	0.000	0.000	PT	0.972	1.024	1.027	1.025
RO	0.03825	0.0455	0.0447	0.0442	RO	0.062	0.067	0.064	0.062	RO	0.099	0.114	0.093	0.052
SK	0.0265	0.02765	0.02745	0.03095	SK	0.265	0.264	0.112	0.089	SK	0.000	0.000	0.000	0.000
SI	0.0375	0.0386	0.03895	0.04345	SI	0.050	0.049	0.047	0.045	SI	0.382	0.426	0.468	0.460
ES	0.04315	0.0411	0.0407	0.0442	ES	0.000	0.000	0.000	0.000	ES	0.091	0.090	0.077	0.067
SE	0.0189	0.0177	0.0166	0.0171	SE	0.018	0.017	0.016	0.016	SE	0.425	0.428	0.426	0.424
UK	0.09875	0.101	NA	NA	UK	0.059	0.059	NA	NA	UK	1.145	1.209	NA	NA

Table 1. Consumption in ddd/1000/day for each antimicrobial class in EEA countries.

J01	17/18	18/19	19/20	20/21	J01A	17/18	18/19	19/20	20/21	J01CA	17/18	18/19	19/20	20/21
AT	NA	NA	10.194	8.809	AT	NA	NA	0.353	0.350	AT	NA	NA	0.789	0.718
BE	22.531	21.833	19.012	17.025	BE	1.911	1.884	1.761	1.689	BE	4.838	4.883	4.047	3.284
BG	20.800	20.910	21.707	22.077	BG	1.639	1.665	1.964	2.326	BG	2.901	2.791	2.712	2.704
HR	18.685	18.793	17.221	16.903	HR	1.022	0.988	0.892	0.866	HR	1.914	1.803	1.484	1.224
CY	28.436	29.055	29.522	26.979	CY	3.657	3.797	3.565	3.481	CY	1.923	1.663	1.385	1.239
CZ	NA	NA	NA	13.540	CZ	NA	NA	NA	1.754	CZ	NA	NA	NA	0.544
DK	15.899	15.427	14.780	14.335	DK	1.454	1.491	1.638	1.727	DK	3.688	3.646	3.563	3.513
EE	11.682	11.774	11.117	10.263	EE	1.299	1.331	1.387	1.375	EE	1.648	1.667	1.503	1.314
FI	15.573	15.050	13.273	11.588	FI	3.330	3.185	2.826	2.447	FI	2.691	2.667	2.238	1.815
FR	25.022	25.216	22.710	20.938	FR	3.102	3.208	3.102	3.010	FR	8.622	8.929	7.712	6.566
DE	21.674	21.251	19.328	17.911	DE	2.345	2.335	2.026	1.739	DE	4.119	4.029	3.451	2.988
GR	34.176	34.110	31.100	25.829	GR	2.924	3.049	3.070	3.037	GR	4.619	4.619	4.275	3.800
HU	14.699	14.632	12.835	11.594	HU	1.121	1.188	1.229	1.241	HU	0.750	0.680	0.535	0.411
IS	20.546	19.835	17.889	16.618	IS	5.221	5.285	5.094	4.740	IS	3.946	4.128	3.729	3.473
IE	21.694	22.618	20.684	18.181	IE	2.747	2.847	3.028	3.018	IE	3.375	3.758	3.481	2.982
IT	21.156	21.545	20.051	17.972	IT	0.529	0.571	0.608	0.639	IT	2.061	2.033	1.785	1.507
LV	13.860	13.842	12.898	11.768	LV	2.252	2.280	2.383	2.514	LV	3.043	2.822	2.341	1.874
LT	16.240	16.105	15.138	13.933	LT	1.421	1.431	1.506	1.504	LT	5.119	4.961	4.374	3.742
LU	22.316	21.608	18.621	15.999	LU	1.594	2.070	1.830	1.601	LU	3.100	3.439	2.974	2.475
MT	21.396	20.461	18.628	16.170	MT	1.694	1.736	2.047	2.224	MT	0.619	0.503	0.451	0.423
NL	9.755	9.605	9.004	8.431	NL	1.983	1.905	1.706	1.530	NL	1.372	1.358	1.175	1.030
NO	15.510	15.095	14.415	13.953	NO	2.748	2.739	2.666	2.553	NO	2.151	2.133	2.040	1.909
PL	24.895	24.015	21.077	19.358	PL	2.395	2.273	2.008	1.790	PL	3.522	3.346	2.677	2.170
PT	18.684	19.182	17.231	15.222	PT	0.857	0.875	0.854	0.862	PT	1.909	2.001	1.646	1.286
RO	24.801	25.432	25.471	25.419	RO	0.898	0.931	0.983	1.053	RO	3.501	3.298	3.041	2.802
SK	20.985	20.671	16.884	15.194	SK	1.702	1.735	1.711	1.722	SK	0.782	0.633	0.466	0.355
SI	13.132	13.080	11.581	10.165	SI	0.512	0.547	0.553	0.548	SI	2.339	2.319	1.848	1.420
ES	26.476	25.547	22.322	19.878	ES	1.488	1.502	1.499	1.534	ES	6.059	5.913	4.990	4.139
SE	12.603	12.117	11.068	10.198	SE	2.422	2.304	2.170	2.027	SE	1.210	1.209	1.157	1.092
UK	19.061	18.470	NA	NA	UK	4.970	4.937	NA	NA	UK	3.462	3.331	NA	NA
J01CE	17/18	18/19	19/20	20/21	J01CF	17/18	18/19	19/20	20/21	J01CR	17/18	18/19	19/20	20/21
AT	NA	NA	0.47865	0.31515	AT	NA	NA	0.029	0.028	AT	NA	NA	3.340	2.949
BE	0.04	0.0396	0.0284	0.0204	BE	0.359	0.372	0.368	0.364	BE	5.433	5.400	4.628	4.075
BG	0.13	0.12625	0.07255	0.01505	BG	0.000	0.000	0.000	0.000	BG	2.530	2.646	2.373	2.130
HR	0.51	0.4852	0.36315	0.24515	HR	0.043	0.048	0.046	0.042	HR	5.822	6.011	5.577	5.237
CY	0.36	0.05745	0.1339	0.1754	CY	0.025	0.028	0.025	0.021	CY	7.095	7.557	7.475	4.643
CZ	NA	NA	NA	0.94935	CZ	NA	NA	NA	0.08595	CZ	NA	NA	NA	2.943
DK	3.89	3.66995	3.2711	2.9866	DK	1.775	1.794	1.753	1.753	DK	1.018	0.966	0.903	0.861
EE	0.19	0.17135	0.141	0.1145	EE	0.123	0.113	0.098	0.094	EE	2.173	2.332	2.281	2.230
FI	1.22	1.22375	1.08875	0.94825	FI	0.161	0.185	0.194	0.201	FI	0.872	0.865	0.751	0.623
FR	0.18	0.1848	0.184	0.1775	FR	0.176	0.163	0.144	0.129	FR	5.028	4.900	4.454	4.173
DE	0.51	0.50125	0.4148	0.3213	DE	0.543	0.523	0.372	0.242	DE	4.507	4.486	4.269	4.120
GR	0.31	0.0327	0.03035	0.04235	GR	0.014	0.016	0.016	0.016	GR	6.513	6.648	5.802	4.830
HU	0.13	0.1183	0.0851	0.0516	HU	0.000	0.000	0.001	0.001	HU	3.972	4.095	3.518	2.957
IS	1.87	1.74915	1.56295	1.37995	IS	1.164	1.158	1.081	1.063	IS	2.783	2.380	1.885	1.647
IE	1.20	1.2331	1.08235	1.27885	IE	1.534	1.542	1.521	1.518	IE	4.323	4.203	3.389	2.648
IT	0.00	0.0011	0.0008	0.0003	IT	0.023	0.019	0.012	0.014	IT	7.334	7.590	7.024	6.247
LV	0.09	0.07285	0.0599	0.0469	LV	0.048	0.053	0.055	0.054	LV	1.862	2.092	2.077	1.974
LT	0.29	0.28055	0.21975	0.13475	LT	0.005	0.000	0.000	0.000	LT	1.877	1.823	1.649	1.801
LU	0.02	0.0127	0.0106	0.0136	LU	0.326	0.202	0.187	0.175	LU	4.746	5.257	4.469	3.808
MT	0.21	0.1853	0.211	0.15	MT	0.364	0.196	0.175	0.166	MT	6.606	6.343	5.938	5.334
NL	0.17	0.144	0.1676	0.1476	NL	0.580	0.588	0.580	0.578	NL	1.077	1.056	0.971	0.905
NO	3.17	3.1575	2.8395	2.473	NO	0.795	0.832	0.856	0.867	NO	0.073	0.082	0.091	0.100
PL	0.34	0.3534	0.26615	0.22665	PL	0.023	0.031	0.035	0.037	PL	3.163	3.151	2.692	2.418
PT	0.03	0.0278	0.02495	0.02245	PT	0.848	0.659	0.400	0.355	PT	6.637	6.952	6.307	5.513
RO	0.60	0.57645	0.5444	0.4154	RO	0.461	0.440	0.401	0.358	RO	6.733	7.179	7.150	7.116
SK	1.07	0.9965	0.74005	0.54005	SK	0.000	0.000	0.000	0.000	SK	3.505	3.613	3.142	2.655
SI	1.61	1.532	1.1214	0.7219	SI	0.286	0.320	0.323	0.306	SI	3.276	3.262	3.031	2.814
ES	0.11	0.11395	0.0944	0.0639	ES	0.272	0.267	0.241	0.217	ES	8.179	7.770	6.732	5.968
SE	3.34	3.14025	2.57685	2.16935	SE	1.800	1.793	1.723	1.662	SE	0.318	0.337	0.337	0.328
UK	0.93	0.8796	NA	NA	UK	1.882	1.846	NA	NA	UK	1.118	1.114	NA	NA
J01DB	17/18	18/19	19/20	20/21	J01DC	17/18	18/19	19/20	20/21	J01DD	17/18	18/19	19/20	20/21
AT	NA	NA	0.40155	0.37505	AT	NA	NA	0.886	0.715	AT	NA	NA	0.261	0.259
BE	0.19	0.1789	0.15235	0.14985	BE	1.233	1.256	1.042	0.837	BE	0.098	0.101	0.098	0.093
BG	0.14	0.1215	0.13845	0.16245	BG	3.432	3.473	3.035	2.691	BG	1.597	1.661	1.836	2.373
HR	0.52	0.4591	0.39105	0.35905	HR	1.904	1.927	1.728	1.577	HR	0.519	0.604	0.635	0.727
CY	0.22	0.1675	0.12565	0.10365	CY	4.445	4.700	4.868	4.286	CY	1.048	1.131	1.153	1.112
CZ	NA	NA	NA	0.0925	CZ	NA	NA	NA	1.5863	CZ	NA	NA	NA	0.181
DK	0.004	0.00345	0.0027	0.0027	DK	0.215	0.184	0.164	0.152	DK	0.030	0.029	0.028	0.026
EE	0.246	0.2467	0.2302	0.2047	EE	1.342	1.317	1.243	1.109	EE	0.054	0.053	0.056	0.058
FI	2.016	1.9312	1.7145	1.5445	FI	0.683	0.712	0.616	0.569	FI	0.086	0.076	0.083	0.080
FR	0.105	0.1048	0.100	0.100	FR	0.328	0.285	0.210	0.144	FR	1.340	1.226	1.041	0.913
DE	0.188	0.17525	0.15575	0.14375	DE	1.276	1.266	1.211	1.086	DE	0.785	0.793	0.771	0.780
GR	0.006	0.00505	0.0032	0.0022	GR	7.781	7.673	6.586	4.950	GR	0.329	0.398	0.418	0.447
HU	0.040	0.0429	0.04185	0.03935	HU	1.902	1.858	1.519	1.157	HU	0.486	0.518	0.502	0.494
IS	0.692	0.6533	0.60755	0.61255	IS	0.031	0.036	0.029	0.016	IS	0.129	0.121	0.121	0.128
IE	0.409	0.4487	0.47295	0.54895	IE	0.738	0.744	0.635	0.455	IE	0.100	0.101	0.091	0.086
IT	0.136	0.1399	0.1329	0.1249	IT	0.205	0.209	0.177	0.132	IT	2.026	2.208	2.056	1.743
LV	0.288	0.25935	0.2358	0.1998	LV	0.493	0.509	0.435	0.353	LV	0.450	0.447	0.469	0.442
LT	0.464	0.4213	0.3509	0.2624	LT	1.414	1.518	1.544	1.517	LT	0.170	0.160	0.143	0.126
LU	0.106	0.09325	0.08035	0.07135	LU	2.554	2.905	2.351	1.776	LU	0.350	0.179	0.183	0.181
MT	0.054	0.06425	0.0612	0.0557	MT	2.270	2.182	1.880	1.317	MT	0.689	0.693	0.659	0.491
NL	0.072	0.07305	0.0668	0.0648	NL	0.086	0.090	0.089	0.081	NL	0.074	0.078	0.085	0.096
NO	0.145	0.142	0.1375	0.137	NO	0.021	0.019	0.017	0.011	NO	0.133	0.129	0.116	0.113
PL	0.237	0.145	0.09195	0.07345	PL	3.580	3.359	3.070	2.350	PL	0.200	0.226	0.263	0.364
PT	0.289	0.24885	0.2074	0.1784	PT	1.234	1.315	1.286	1.238	PT	0.415	0.377	0.327	0.339
RO	0.203	0.1734	0.1289	0.1024	RO	3.485	3.615	3.336	2.939	RO	1.319	1.426	1.395	1.430
SK	0.256	0.249	0.19975	0.15125	SK	3.937	3.998	3.371	2.776	SK	1.616	1.689	0.860	0.863
SI	0.136	0.1423	0.12885	0.11085	SI	0.301	0.301	0.366	0.431	SI	0.145	0.155	0.137	0.120
ES	0.145	0.1473	0.14245	0.14395	ES	1.763	1.776	1.590	1.423	ES	0.698	0.726	0.704	0.676
SE	0.077	0.06735	0.0583	0.0543	SE	0.010	0.010	0.010	0.010	SE	0.113	0.112	0.107	0.106
UK	0.216	0.20195	NA	NA	UK	0.043	0.042	NA	NA	UK	0.069	0.073	NA	NA
J01DH	17/18	18/19	19/20	20/21	J01E	17/18	18/19	19/20	20/21	J01FA	17/18	18/19	19/20	20/21
AT	NA	NA	0.10945	0.10945	AT	NA	NA	0.267	0.270	AT	NA	NA	1.293	0.950
BE	0.04925	0.0482	0.046	0.0445	BE	0.236	0.254	0.262	0.269	BE	3.167	3.196	2.755	2.349
BG	0.0295	0.036	0.06985	0.11035	BG	0.862	0.826	0.715	0.650	BG	3.232	3.299	4.124	4.906
HR	0.07515	0.07845	0.08535	0.10285	HR	0.562	0.532	0.499	0.480	HR	2.535	2.546	2.355	2.478
CY	0.13075	0.14005	0.14505	0.16955	CY	0.276	0.228	0.239	0.266	CY	2.851	2.968	2.777	2.573
CZ	NA	NA	NA	0.08925	CZ	NA	NA	NA	0.7719	CZ	NA	NA	NA	2.380
DK	0.0546	0.05395	0.0536	0.0521	DK	0.806	0.750	0.692	0.669	DK	1.608	1.520	1.370	1.210
EE	0.05635	0.06235	0.0726	0.0796	EE	0.474	0.469	0.396	0.346	EE	2.279	2.313	2.058	1.749
FI	0.05465	0.04695	0.0414	0.0424	FI	1.084	1.111	1.095	1.029	FI	0.577	0.523	0.405	0.293
FR	0.03375	0.03325	0.032	0.0345	FR	0.472	0.470	0.494	0.520	FR	1.943	1.832	1.662	1.563
DE	0.0501	0.0525	0.0569	0.0639	DE	0.659	0.657	0.632	0.596	DE	2.951	2.881	2.571	2.366
GR	0.16865	0.15855	0.16435	0.19385	GR	0.329	0.332	0.403	0.431	GR	6.391	6.180	5.584	4.133
HU	0.04705	0.05015	0.05815	0.06665	HU	0.474	0.474	0.444	0.410	HU	2.339	2.338	2.094	2.183
IS	0.0242	0.03295	0.0315	0.0345	IS	0.551	0.570	0.554	0.542	IS	1.601	1.479	1.255	1.078
IE	0.04635	0.04625	0.0463	0.0468	IE	0.993	1.076	1.090	1.089	IE	4.247	4.079	3.498	2.787
IT	0.0381	0.0474	0.05435	0.05885	IT	0.816	0.934	1.033	1.046	IT	3.962	4.122	4.036	3.615
LV	0.03135	0.03235	0.0426	0.0461	LV	0.863	0.836	0.700	0.533	LV	2.052	2.128	1.896	1.630
LT	0.03935	0.0417	0.04655	0.05405	LT	0.688	0.770	0.748	0.554	LT	2.139	2.159	1.956	1.639
LU	0.05725	0.05285	0.04985	0.04985	LU	0.451	0.319	0.339	0.347	LU	3.714	2.987	2.533	2.099
MT	0.15435	0.08845	0.08465	0.08765	MT	0.410	0.407	0.399	0.386	MT	4.518	4.337	3.365	2.380
NL	0.01505	0.01485	0.0148	0.0148	NL	0.454	0.470	0.480	0.478	NL	1.224	1.247	1.200	1.122
NO	0.02625	0.0247	0.02395	0.02445	NO	0.728	0.764	0.775	0.765	NO	0.791	0.723	0.610	0.468
PL	0.0244	0.0284	0.03365	0.03965	PL	0.552	0.575	0.514	0.429	PL	4.507	4.250	2.687	2.571
PT	0.08185	0.07795	0.0781	0.0856	PT	0.446	0.424	0.428	0.429	PT	2.645	2.985	2.667	2.053
RO	0.0572	0.06745	0.08095	0.10095	RO	0.855	0.843	0.798	0.742	RO	2.773	2.883	3.721	4.515
SK	0.0425	0.0449	0.04495	0.05245	SK	0.406	0.445	0.462	0.451	SK	4.286	4.222	3.535	3.426
SI	0.05845	0.05415	0.053	0.058	SI	0.741	0.713	0.630	0.583	SI	1.605	1.628	1.412	1.182
ES	0.0877	0.0866	0.08735	0.09385	ES	0.467	0.465	0.454	0.458	ES	2.976	2.848	2.386	1.997
SE	0.0394	0.0403	0.04045	0.04145	SE	0.313	0.322	0.320	0.312	SE	0.239	0.243	0.271	0.265
UK	0.0524	0.0504	NA	NA	UK	1.069	0.942	NA	NA	UK	2.977	2.772	NA	NA
J01FF	17/18	18/19	19/20	20/21	J01G	17/18	18/19	19/20	20/21	J01M	17/18	18/19	19/20	20/21
AT	NA	NA	0.62585	0.58085	AT	NA	NA	0.028	0.026	AT	NA	NA	0.799	0.683
BE	0.4319	0.4505	0.43235	0.40235	BE	0.034	0.034	0.033	0.031	BE	1.847	1.031	0.662	0.590
BG	0.79515	0.82685	0.913	0.8825	BG	0.254	0.256	0.256	0.255	BG	2.996	2.963	3.303	3.820
HR	0.41405	0.43615	0.44415	0.46515	HR	0.098	0.095	0.081	0.075	HR	1.727	1.662	1.509	1.482
CY	0.11645	0.1413	0.122	0.1225	CY	0.078	0.079	0.084	0.091	CY	5.751	5.558	6.580	6.679
CZ	NA	NA	NA	0.40275	CZ	NA	NA	NA	0.081	CZ	NA	NA	NA	0.596
DK	0.06945	0.0714	0.07505	0.08005	DK	0.056	0.057	0.053	0.047	DK	0.565	0.519	0.465	0.433
EE	0.177	0.1761	0.1723	0.1668	EE	0.037	0.028	0.024	0.024	EE	0.924	0.817	0.828	0.838
FI	0.27765	0.27225	0.24955	0.23155	FI	0.015	0.017	0.016	0.015	FI	0.854	0.720	0.584	0.533
FR	0.1276	0.14795	0.1668	0.1768	FR	0.076	0.070	0.064	0.059	FR	1.543	1.452	1.311	1.225
DE	0.24335	0.25425	0.2683	0.2778	DE	0.090	0.086	0.076	0.068	DE	1.890	1.710	1.536	1.501
GR	0.4158	0.44065	0.41455	0.34855	GR	0.139	0.142	0.134	0.126	GR	2.979	3.197	3.019	2.691
HU	0.5974	0.596	0.5788	0.5608	HU	0.032	0.028	0.033	0.033	HU	2.554	2.285	1.808	1.544
IS	0.21445	0.19975	0.17445	0.16445	IS	0.026	0.026	0.028	0.029	IS	0.909	0.761	0.574	0.523
IE	0.1035	0.1024	0.1003	0.0973	IE	0.101	0.103	0.091	0.077	IE	0.884	0.745	0.552	0.465
IT	0.0312	0.03375	0.0339	0.0314	IT	0.068	0.063	0.057	0.056	IT	3.043	2.628	2.059	1.848
LV	0.14955	0.15445	0.15755	0.15555	LV	0.076	0.069	0.063	0.056	LV	1.249	1.148	0.998	0.898
LT	0.02825	0.02675	0.02495	0.02345	LT	0.069	0.065	0.057	0.049	LT	1.086	1.042	0.972	0.885
LU	0.89745	0.5109	0.47785	0.45385	LU	0.053	0.044	0.032	0.022	LU	2.585	1.963	1.551	1.384
MT	0.18435	0.21595	0.2191	0.2351	MT	0.081	0.089	0.076	0.057	MT	2.540	2.360	1.879	1.595
NL	0.246	0.2546	0.2562	0.2592	NL	0.055	0.052	0.047	0.043	NL	0.816	0.777	0.723	0.703
NO	0.22275	0.2192	0.20965	0.19765	NO	0.087	0.093	0.090	0.084	NO	0.380	0.338	0.293	0.265
PL	0.8168	0.80095	0.74665	0.73715	PL	0.078	0.074	0.065	0.058	PL	1.660	1.588	1.406	1.328
PT	0.0739	0.0771	0.0759	0.0809	PT	0.060	0.058	0.059	0.060	PT	1.839	1.728	1.466	1.286
RO	0.17485	0.19395	0.20495	0.20345	RO	0.204	0.191	0.172	0.154	RO	3.274	3.310	3.223	3.228
SK	0.5469	0.56615	0.5575	0.5365	SK	0.103	0.102	0.077	0.072	SK	2.371	2.003	1.490	1.327
SI	0.2569	0.25855	0.23775	0.21875	SI	0.061	0.061	0.058	0.054	SI	1.305	1.234	1.069	0.973
ES	0.21775	0.21625	0.1997	0.2062	ES	0.081	0.079	0.077	0.078	ES	3.042	2.746	2.250	1.955
SE	0.34235	0.32515	0.29815	0.27315	SE	0.026	0.024	0.023	0.022	SE	0.762	0.721	0.659	0.623
UK	0.100	0.100	NA	NA	UK	0.129	0.132	NA	NA	UK	0.568	0.542	NA	NA
J01XA	17/18	18/19	19/20	20/21	J01XD	17/18	18/19	19/20	20/21	J01XE	17/18	18/19	19/20	20/21
AT	NA	NA	0.02215	0.02365	AT	NA	NA	0.056	0.0552	AT	NA	NA	0.272	0.229
BE	0.042	0.0415	0.0407	0.0422	BE	0.020	0.020	0.019	0.018	BE	2.373	2.403	2.398	2.354
BG	0.02415	0.02355	0.017	0.021	BG	0.136	0.140	0.114	0.080	BG	0.000	0.000	0.000	0.000
HR	0.04565	0.04855	0.0482	0.0577	HR	0.129	0.133	0.110	0.113	HR	0.735	0.813	0.845	0.874
CY	0.085	0.09665	0.1032	0.1032	CY	0.196	0.242	0.254	0.191	CY	0.443	0.460	0.439	0.408
CZ	NA	NA	NA	0.0321	CZ	NA	NA	NA	0.067	CZ	NA	NA	NA	0.912
DK	0.02485	0.0246	0.0248	0.0248	DK	0.087	0.081	0.075	0.070	DK	0.215	0.217	0.274	0.278
EE	0.02235	0.02385	0.02385	0.02635	EE	0.044	0.042	0.040	0.035	EE	0.573	0.582	0.526	0.495
FI	0.0313	0.02735	0.0239	0.0269	FI	0.032	0.027	0.029	0.049	FI	0.405	0.310	0.291	0.314
FR	0.0371	0.0355	0.03315	0.03315	FR	0.054	0.052	0.050	0.051	FR	0.209	0.177	0.140	0.117
DE	0.05065	0.04965	0.0434	0.0404	DE	0.053	0.055	0.051	0.048	DE	0.811	0.824	0.845	0.805
GR	0.14485	0.1488	0.1518	0.1633	GR	0.187	0.169	0.147	0.124	GR	0.656	0.601	0.550	0.566
HU	0.0228	0.0216	0.02195	0.02445	HU	0.043	0.045	0.059	0.071	HU	0.085	0.183	0.193	0.191
IS	0.0322	0.03195	0.0171	0.0021	IS	0.054	0.052	0.051	0.052	IS	0.682	0.515	0.492	0.497
IE	0.0754	0.0773	0.074	0.068	IE	0.056	0.057	0.050	0.044	IE	0.691	1.369	1.383	1.376
IT	0.0602	0.05415	0.04905	0.04505	IT	0.031	0.045	0.047	0.041	IT	0.191	0.205	0.222	0.236
LV	0.0195	0.0309	0.04045	0.03295	LV	0.144	0.134	0.130	0.120	LV	0.711	0.731	0.769	0.807
LT	0.035	0.03895	0.04645	0.04745	LT	0.089	0.101	0.097	0.057	LT	1.222	1.279	1.403	1.439
LU	0.0382	0.03255	0.0315	0.0325	LU	0.037	0.038	0.036	0.033	LU	0.885	1.293	1.281	1.268
MT	0.12625	0.13565	0.14965	0.16315	MT	0.109	0.112	0.098	0.083	MT	0.428	0.490	0.697	0.811
NL	0.01995	0.01985	0.02025	0.02075	NL	0.034	0.034	0.033	0.032	NL	1.371	1.343	1.287	1.257
NO	0.01495	0.01495	0.01475	0.01425	NO	0.039	0.039	0.038	0.035	NO	0.244	0.214	0.218	0.223
PL	0.01825	0.0198	0.02265	0.03615	PL	0.072	0.078	0.079	0.083	PL	3.634	3.633	4.336	4.349
PT	0.04605	0.03855	0.0382	0.0372	PT	0.025	0.000	0.000	0.000	PT	0.972	1.024	1.027	1.025
RO	0.03825	0.0455	0.0447	0.0442	RO	0.062	0.067	0.064	0.062	RO	0.099	0.114	0.093	0.052
SK	0.0265	0.02765	0.02745	0.03095	SK	0.265	0.264	0.112	0.089	SK	0.000	0.000	0.000	0.000
SI	0.0375	0.0386	0.03895	0.04345	SI	0.050	0.049	0.047	0.045	SI	0.382	0.426	0.468	0.460
ES	0.04315	0.0411	0.0407	0.0442	ES	0.000	0.000	0.000	0.000	ES	0.091	0.090	0.077	0.067
SE	0.0189	0.0177	0.0166	0.0171	SE	0.018	0.017	0.016	0.016	SE	0.425	0.428	0.426	0.424
UK	0.09875	0.101	NA	NA	UK	0.059	0.059	NA	NA	UK	1.145	1.209	NA	NA

3.1. Associations Between Overall Antibiotic Use and Resistance

Strong associations were detected between all 12 resistotypes and overall antibiotic consumption on univariate analysis with effect sizes ranging from 0.392 for VRE to 0.666 for MRSA (Table 2). In multivariate analyses this remained significant for 3GCR and 3XR in E. coli, and for 3GCR, AGR and 3XR in K. pneumoniae (Table 3). At the beginning of the analysed period in 2017-2018, overall antibiotic consumption varied from a low of 9.755 ddd/1000/day in the Netherlands to a high of 34.176 ddd/1000/day in Greece. This decreased over time to a low of 8.431 ddd/1000/day in the Netherlands to a high of 26.979 ddd/1000/day in Cyprus for season 2020-2021. This >3-fold disparity in antibiotic consumption may imply that there is much scope for countries with high consumption to reduce use. Indeed, Greece, the country with highest consumption at outset of analysis had decreased overall consumption by approximately one-third by the conclusion of the analysis period, from 34.176 ddd/1000/day to 25.829 ddd/1000/day.

3.2. Associations Between Tetracycline Use and Resistance

Mean J01A use differed little between group 1 and 2 countries (Figure 7). Usage of tetracyclines had weakly negative correlations on univariate analysis with all resistotypes other than VRE, for which the association was weakly positive albeit statistically insignificant (Table 2). The weakly negative correlations found on univariate analysis for tetracycline use were significant only in the case of 3GCR, FQR and 3XR in K. pneumoniae and for MRSA (Table 2). In multivariate analysis, the weakly negative associations with AMPR in E. coli and FQR in K. pneumoniae were the only correlations found to be independently significant (Table 3). It is not clear why tetracyclines, having a substantial Gram-negative spectrum, appear to be inversely associated with resistance in E. coli or K. pneumoniae. Determinants of tetracycline resistance have long been known to co-localise with other resistance genes on mobile genetic elements and spread readily amongst human and animal hosts, amplified by co-selection from tetracyclines and other agents [75,76,77,78,79]. A possible explanation is that these primarily bacteriostatic agents exert, relatively speaking, a weaker selective pressure than alternative drugs. In other words, the relationship is determined by what tetracyclines are being used instead of rather than by lack of selective pressure from tetracyclines per se. The low resistance potential of extended spectrum penicillins (J01CA), 1st generation cephalosporins (J01DB) and sulphonamides/trimethoprim (J01E), all intrinsically active against coliform organisms to some extent, might be similarly explained. Two recent studies of travellers to tropical areas with very high prevalence of ESBL producing Enterobacterales demonstrated that daily use of 100mg of doxycycline orally for malaria prophylaxis was not an independent risk factor for gut colonisation by these organisms whereas use of other antibiotics including fluoroquinolones, macrolides and β-lactams was an independent risk factor [80,81,82]. Another study recently evaluated the use of doxycycline postexposure prophylaxis (PEP) for bacterial sexually transmitted infections and found no increase in ESBL carriage amongst the study population though tetracycline resistance did increase amongst incident gonorrhoea cases [83,84]. The majority of J01A use in the EEA comprises doxycycline which may skew these results, and it is quite possible that other, lesser used tetracyclines differ from doxycycline in resistance potential [74]. Indeed, it has previously been found that doxycycline selects less readily for resistance amongst commensal E. coli than does tetracycline itself, presumably because the latter is less completely absorbed from the gut lumen with consequently greater exposure to the mucosal flora [85]. The results obtained here support the inclusion of doxycycline, if not other tetracyclines, in the WHO access category [1,2,3,4,5]. Recent evidence indicates that doxycycline compares favourably to other agents in mild to moderate hospital acquired pneumonia (HAP) and may spare the use of agents with higher resistance potential such as cefepime, piperacillin/tazobactam and carbapenems, allowing them to be reserved for moderate to severe HAP where multidrug resistant Gram-negative organisms are more likely to be implicated and stakes from early clinical failure are unacceptably high [44,86]. At least in areas with low pneumococcal resistance, doxycycline is a useful alternative agent to penicillins for community acquired pneumonia (CAP) and may be preferable to macrolides in this regard given its better coverage of pneumococci and Haemophilus influenzae, lower resistance potential, low risks for cytochrome P450 mediated drug interactions and lack of torsadogenic QTc prolongation [43,45,47,49,87]. Moreover, in atypical pneumonia, coverage of Mycoplasma pneumoniae and Chlamydophila spp. from doxycycline is comparable to that of macrolides and likely greater where macrolide resistant M. pneumoniae is endemic [87,88]. Although active against Legionella pneumophila, doxycycline is less potent against this organism than are newer macrolides or fluoroquinolones [89]. Given the high morbidity and mortality associated with Legionnaires’ disease, macrolides or levofloxacin should probably be prescribed preferentially in atypical pneumonia where this is a suspected or proven aetiology [89,90]. If non-pneumophila Legionella spp. is involved this is even more critical given that some species, notably L. longbeachae, are inherently resistant to tetracyclines [91]. As tetracyclines remain active against some MDR-GNB, interest in their use for urinary, respiratory and other infections by these organisms has been rekindled over the last decade. Several reports of clinical success with unorthodox use of doxycycline and minocycline in such miscellaneous infections have been published [92,93]. There have even been instances where doxycycline was used successfully in urinary tract infection (UTI) caused by Pseudomonas aeruginosa, a pathogen intrinsically resistant to tetracyclines [93]. This highlights the underappreciated fact that susceptibilities are typically reported by laboratories based upon concentrations that are readily attainable in serum [93]. Tetracyclines and many other antibiotics undergoing renal elimination, penicillins included, may reach peak concentrations in urine ≥2 orders of magnitude greater than in serum, overpowering ‘resistant’ pathogens in uncomplicated UTI which does not involve the upper urinary tract or prostate [94,95,96]. Consideration of lower urinary breakpoints may have the potential to broaden treatment options for these common infections and thereby improve antimicrobial stewardship [94,95,96].

3.3. Associations Between Penicillin Use and Resistance

Extended spectrum penicillin (J01CA) use was high in most EEA countries (Table 1). In group 1 countries, J01CA consumption fell from 3.671 to 2.943 and in group 2 countries from 3.492 to 2.583 ddd/1000/day (Figure 5). Consumption of this group of drugs, which includes the aminopenicillins ampicillin and amoxicillin, did not correlate positively with any resistotype in univariate analysis at the α-level of 0.05 (Table 2). On multivariate analysis, extended spectrum penicillin use showed weak negative associations with MRSA, VRE and FQR E. coli and a weak positive association with AMPR E. coli, all independently significant (Table 3). The fact that the effect size was small might reflect the fact that all countries had both high consumption of these agents (Table 1) and exceedingly high levels of AMPR in E. coli (Figure 1). The finding that aminopenicillin use was not an independent predictor of any other resistotype supports the assignment of these drugs to the WHO access category [1,2,3,4,5]. Nevertheless, aminopenicillin consumption has previously been documented as a risk factor for emergent resistance not only to aminopenicillins themselves but also to staple drugs for UTI treatment, including trimethoprim, amongst uropathogenic E. coli [96,97,98,99]. Their resistance potential is higher than that of β-lactamase labile penicillins (J01CE), which include benzylpenicillin (penicillin G) and phenoxymethylpenicillin (penicillin V). These narrower spectrum agents, unlike aminopenicillins, were not associated with AMPR in E. coli, or any resistotype, in univariate or multivariate analysis (Table 2 and Table 3). In group 1 countries, use of J01CE drugs declined from 1.264 to 1.028 and in group 2 countries from 0.337 to 0.169 ddd/1000/day (Figure 4). Penicillins G or V, it may be argued, should be used preferentially where Gram-negative cover is unnecessary. Amongst the countries with the lowest levels of AMPR E. coli, were the Nordic nations which had high utilisation of β-lactamase labile penicillins (Table 1). In contrast to other EEA nations, they continue to use penicillins G and V, rather than amoxicillin, as first line therapy for community acquired pneumonia (CAP), pharyngotonsillitis, otitis media (OM) and dentoalveolar infections [100,101]. It has been suggested that these drugs exert less selective pressure than aminopenicillins for resistance in Enterobacterales and should be used in preference to them whenever possible [100,101]. Detractors from this position argue that amoxicillin has the advantages of activity against H. influenzae, alongside greater bioavailability and palatability when given by the oral route [101,102]. Rhedin and colleagues assessed outcomes in CAP amongst Swedish children aged 1 to 5 years using penicillin V versus amoxicillin and found treatment failures were significantly higher with penicillin V (7.7 %) versus amoxicillin (4.7 %) [102]. Nevertheless, there was no difference in the incidence of serious complications or mortality between the groups and the number needed to treat with amoxicillin to prevent one clinical failure was quite large at 31 [102]. The same group demonstrated noninferiority of oral penicillin V in adult CAP patients with CRB-65 score ≤1 and of IV penicillin G in those with CRB-65 score of 2 [103]. These studies indicate that the longer-term ecological benefits of using penicillins G & V likely outweigh modest clinical gains from using amoxicillin and suggest that the causal role of H. influenzae in CAP amongst otherwise healthy patients is minor, at least where immunisation rates against virulent capsular serogroup B strains are high and pneumococcal resistance to penicillin is low [102,103]. It is possible that oral penicillin V may be at a disadvantage when compared to oral amoxicillin in countries where penicillin resistant pneumococci are more commonly encountered than in Scandinavia [104]. One study attempted to address this hypothesis in Spain, a country with a high prevalence of penicillin resistance and found penicillin V inferior in intention to treat but not on per protocol analyses [104]. Those investigators, however, cautioned that their study had been underpowered as they struggled to recruit an adequately large sample cohort [104]. The activity of penicillin G against H. influenzae, though weaker than that of aminopenicillins, is greater than that of penicillin V. Thegerstrӧm and coworkers found that intravenous (IV) penicillin G did not achieve worse outcomes than IV aminopenicillins in pneumonia cases from which H. influenzae was isolated, although noting a trend towards slower clinical response with penicillin G [105]. Subsets of patients with chronic obstructive pulmonary disease (COPD) or bronchiectasis, who are known to be especially susceptible to colonisation and infection with H. influenzae, it might be argued, should receive targeted treatment against this organism [106]. A recent study in the UK found, however, that aminopenicillin resistance in H. influenzae isolates from COPD patients was disproportionately higher (67 %) than those from the wider UK population at 20-25 % [106]. COPD patients are also liable to infection with various other organisms including Moraxella catarrhalis, now invariably resistant to aminopenicillins [107]. The Gram-negative spectrum of aminopenicillins has already been much eroded by acquired resistance, but amoxicillin is still an option for definitive treatment of infections caused by E. coli, Proteus, Salmonella and H. influenzae strains with laboratory proven sensitivity [108]. It is difficult to envisage a scenario where IV amoxicillin would be broadly superior to IV penicillin G in empirical therapy. When either agent is blindly chosen for undifferentiated infection nowadays, it is with the intention of covering Gram-positive pathogens, a purpose for which amoxicillin has a needlessly broad spectrum. Use of penicillin G for streptococcal/enterococcal cover could therefore be more appropriate, with an aminoglycoside added, if necessary, for aerobic Gram-negative coverage. Much has been made of the fact that Enterococci have lower minimum inhibitory concentrations (MICs) for aminopenicillins than for penicillin G [108,109]. However, the absolute difference in activity is small, equating to approximately 1 doubling dilution, and has never been convincingly shown to have any clinical impact at least for ‘wild type’ strains with MICs below the epidemiologic cut-off [108,109,110,111]. Penicillin G also has the advantage of being more stable than aminopenicillins in solution, making it potentially more convenient to administer, particularly in the contexts of using prolonged/continuous infusion to maximise time-dependent bactericidality or in outpatient parenteral antimicrobial therapy (OPAT) [111,112]. Misleadingly, mecillinam (amdinocillin) and its orally administered pivaloyl ester, pivmecillinam, are grouped with the aminopenicillins under ATCC code J01CA, extended spectrum penicillins, as is temocillin [74]. Both mecillinam and temocillin differ substantially from aminopenicillins, not least by their near total lack of activity against Gram-positive organisms and obligate anaerobes [113,114,115,116,117,118]. Each has a spectrum of action limited almost exclusively to Enterobacterales with little cross-resistance to aminopenicillins [113,114,115,116,117,118]. Pivmecillinam has been established as frontline treatment for uncomplicated bacterial cystitis in Nordic countries since the 1970s with minimal associated resistance and accounts for a substantial proportion of J01CA consumption there [119]. Though not an officially approved indication, there is some data to suggest that when adequately dosed, mecillinam may concentrate sufficiently in the renal parenchyma to be effective even in upper UTI with bacteraemic overspill [120,121]. Elsewhere, J01CA usage is almost entirely comprised of aminopenicillins, hence resistance potential for mecillinam and temocillin cannot be inferred from these data [74]. This is unfortunate since both agents have several ideal characteristics making them warrant investigation for wider applicability in invasive enterobacterial infections. These include bactericidality, low rates of resistance in target Gram-negative pathogens, minimal toxicity, a narrow spectrum conferring lower propensity to distort the gut flora than many comparable agents and in the case of mecillinam, potential for frequent synergy with other β-lactams via complementary binding of different transpeptidase targets [113,118,119,122,123]. Use of β-lactamase stable penicillins (J01CF), chiefly comprising the narrow spectrum antistaphylococcal penicillins (cloxacillin, flucloxacillin and dicloxacillin) was not associated with any resistotype on univariate analysis (Table 2). Use of β-lactamase stable penicillins (J01CF) had independently negative correlations with MRSA and with 3GCR, AGR, FQR and 3XR in K. pneumoniae on multivariate analysis (Table 3). This may reflect unavailability or prohibitively inflated costs in some EEA countries including Bulgaria, Hungary, Lithuania and Slovakia (Table 1). As a result, broad-spectrum alternatives such as cephalosporins (J01D) and penicillin/β-lactamase inhibitor combinations (J01CR) may have been substituted for skin and soft tissue infections (SSTI) with resultant increases in collateral resistance. Lending credence to this argument, use of J01CF agents was 6-to-7-fold lower in group 2 than in group 1 countries (Figure 5). Use of prolonged or continuous infusions via elastomeric devices and coadministration of probenecid with both oral and IV formulations have been proposed as methods which may improve the pharmacokinetic profile of various narrow-spectrum penicillins in deep seated infections requiring prolonged high dose treatment [111,112,124,125,126]. Such strategies may obviate the need to use longer acting but much broader spectrum agents, such as ceftriaxone, in OPAT for deep seated staphylococcal and streptococcal infections. Consumption of penicillin/β-lactamase inhibitor (J01CR) agents was found to be high in most EEA countries and in many of these, surpassed the use of narrower spectrum penicillins belonging to the J01CA, J01CE and J01CF groups (Table 1). Mean baseline consumption of these drugs was over twice as high in group 2 countries at 5.72 ddd/1000/day than in group 1 countries (2.73 ddd/1000/day) in 2017-2018 (Figure 5). Although both groups had reduced consumption by 2020-2021, this was by <20 % in each case (Figure 5). Use of penicillin/β-lactamase inhibitor combinations (J01CR) had positive associations of varying strength with all 12 resistotypes in univariate analysis (Table 2) but independent significance on multivariate analysis held only for MRSA, CARBR in K. pneumoniae, FQR and AMPR in E. coli (Table 3). Extended spectrum penicillin/β-lactamase inhibitor combinations were more strongly associated with AMPR than were unpotentiated aminopenicillins (Table 2 and Table 3). This may result from selection of strains hyperproducing penicillinases such as TEM-1 [127,128]. Inhibitor combinations may, paradoxically, exert stronger selective pressure than unprotected penicillins if only strains expressing exceedingly high levels of β-lactamase sufficient to overcome enzymatic inhibition gain a survival advantage [127,128]. As has been reported previously, co-resistance to fluoroquinolones may be present in such strains, potentially explaining the strong association observed between FQR in E. coli and use of class J01CR agents [129,130]. Carbapenem resistant Enterobacterales have high MICs for extended spectrum penicillin/β-lactamase inhibitor combinations, generally higher than for carbapenems themselves [131]. This could explain the relationship found between CARBR in K. pneumoniae and consumption of these drugs (Table 3). The strong link between MRSA and use of J01CR agents was not expected. Whilst it seems intuitive that such drugs would select for MRSA, the strength of this association was not expected to be so great given that preferential substitution of these agents for cephalosporins and fluoroquinolones in hospital formularies has been apparently successful in reducing the incidence of nosocomial MRSA infections [131,132,133]. There have been, however, several previous studies detecting positive associations between extended spectrum penicillin/β-lactamase inhibitor combination usage and incidence of colonisation or infection by MRSA [134]. Furthermore, these agents were heavily used throughout the EEA and were the most used systemic antibacterials in many countries (Table 1), demonstrating that excessive use of any agent can generate resistance. Heavy consumption of these broad-spectrum agents presents an obvious though difficult target for stewardship initiatives as there are not many clear alternatives for empirical treatment of certain serious infections, such as HAP and intraabdominal sepsis, which have diverse microbiological aetiologies. Nevertheless, wide discordance in use throughout the EEA (Figure 5) would imply that sizeable reductions in usage are theoretically achievable, at least for countries with heavier baseline consumption. Ironically, increased reliance on extended spectrum penicillin/β-lactamase inhibitors in some countries, including the UK, resulted from earlier admonishments to curtail prescribing of expanded spectrum cephalosporins and fluoroquinolones [131]. Whilst use of broad-spectrum agents such as piperacillin-tazobactam will be justified in many HAP cases, evidence suggests that a subset of HAP patients do not require broad Gram-negative cover [44,86]. For intrabdominal sepsis use of J01CR agents will often be appropriate but a combination of an aminoglycoside and penicillin G (with or without metronidazole based on risk for anaerobic involvement) might be a reasonable alternative in most cases [135,136,137,138].

3.4. Associations Between Cephalosporin Use and Resistance

In group 1 countries, 1st generation cephalosporin (J01DB) use was 0.167 and 0.173 ddd/1000/day in 2017/2018 and 2020/2021, and for group 2 countries, 0.21 and 0.135 ddd/1000/day, respectively (Figure 6). Practically all countries had low consumption of these drugs (Table 1). The sole exception was Finland, a country with some of the lowest resistance levels in the EEA (Figure 1, Figure 2 and Figure 3). Use of 1st generation cephalosporins was not positively associated with any resistotype on univariate analysis (Table 2). In multivariate analysis J01DB consumption had significant negative correlations with VRE, AMPR E. coli and with CARBR, 3GCR and FQR, but not AGR or 3XR, in K. pneumoniae (Table 3). This suggests that 1st generation cephalosporins, unlike 2nd and 3rd generation analogues, have comparatively low resistance potential. This conclusion has also been drawn elsewhere from a recent meta-analysis of data collected at multinational level [40]. Accordingly, allocation of many of these J01DB agents to the WHO access group seems valid and they merit consideration for management of infections caused by E. coli, Klebsiella and Proteus strains of established or strongly suspected susceptibility [1,2,3,4,5]. Examples would include use of agents such as cefazolin in pyelonephritis or biliary tract infection as definitive treatment or follow-on treatment after initial use of an IV aminoglycoside whilst sensitivities are awaited [135,136,137,138,139,140]. Though 1st generation cephalosporins have less resistance potential than do their 2nd or 3rd generation counterparts, they still have a substantial Gram-negative spectrum which should not be squandered on uncomplicated SSTIs such as cellulitis and infection of ‘clean’ wounds caused by Gram-positive cocci, where antistaphylococcal penicillins should suffice, assuming lack of specific risk factors for involvement of MRSA or Gram-negative organisms [135,139,141,142,143]. In patients with genuine penicillin allergies, cautious use of specific J01DB congeners with R1 side chains lacking cross-reactivity e.g., cefazolin, may be justified in some circumstances [144]. Nitrofurantoin and pivmecillinam are probably better options for treating simple cystitis given that these drugs are not, unlike 1st generation cephalosporins, deemed to be useful in systemic infection [96,97,118,119]. Consumption of 2nd generation cephalosporins (J01DC) was high in many countries (Table 1) but declined over the study period in both groups 1 and 2, from 0.566 to 0.399 and from 3.77 to 2.888 ddd/1000/day, respectively (Figure 6). Mean 3rd generation cephalosporin (J01DD) usage in group 2 countries increased from 0.816 to 1.018 ddd/1000/day but fell from 0.261 to 0.198 ddd/1000/day over the same time for group 1(Figure 6). Six of seven group 2 countries had increasing J01DD use, consumption in the remaining country, Spain, was stable (Table 1). Both 2nd and 3rd generation cephalosporins were strongly associated with all 12 resistotypes on univariate analysis (Table 2). For 2nd generation agents, these strong associations remained independently significant on multivariate analysis for MRSA, VRE and FQR in E. coli and all resistotypes in K. pneumoniae (Table 3). In the case of 3rd generation drugs significance on multivariate analysis was maintained for all resistotypes other than AMPR in E. coli and, unexpectedly, for VRE (Table 3). Profligate use of 2nd and 3rd generation cephalosporins poses a further target for antimicrobial policymakers as these agents are being used in clinical situations where narrower spectrum agents with less resistance potential would be more appropriate. The wide disparity in use between countries implies that this should not be an unrealistic undertaking. Based on the analysis presented here, it cannot be determined whether individual 2nd and 3rd generation cephalosporins differ in terms of resistance potential and it is possible that this is the case even though these classes have an overall high risk for resistance selection [26,27,28,29,30,132,133]. Some studies have found that cefotaxime, which undergoes much less biliary excretion than ceftriaxone, exerts less selective pressure upon the bowel flora for C. difficile and MDR-GNB though other studies have yielded contradictory findings [145,146,147,148].

3.5. Associations Between Carbapenem Use and Resistance

Carbapenem (J01DH) use was stable in group 1 countries at ~0.04 ddd/1000/day over 5 years but increased by almost one-third in group 2 countries in the same period from a baseline of 0.0819 ddd/1000/day, already more than double that of group 1 countries (Figure 8). Not surprisingly, carbapenem use correlated strongly with CARBR, 3GCR and FQR in K. pneumoniae on multivariate analysis (Table 3). It seems probable that the relationship between carbapenem consumption and CARBR is causal, and conversely, that increased 3GCR and FQR will have fuelled reliance on carbapenems as has been reported previously [149]. Worryingly, resistance to this crucial class of ‘last resort’ antimicrobials was high and rising in many countries (Figure 2). By 2021, almost 75 % of K. pneumoniae isolates from Greece and over half from Romania were carbapenem resistant [73].

3.6. Associations Between Sulphonamide/Trimethoprim Use and Resistance

Consumption of sulphonamides/trimethoprim (J01E) differed little between group 1 and 2 countries, remaining stable in both over 5 years (Figure 7). It had a very weak, yet significant, positive association with VRE and with 3GCR in E. coli but had a significant negative relationship of similar magnitude with combined resistance (3XR) in E. coli, on multivariate analysis (Table 3). Significant albeit weak negative associations were also evident between J01E use and MRSA as well as for 3GCR, FQR and CARBR resistotypes in K. pneumoniae (Table 3). Overall, this suggests that these agents have a low resistance potential and that their inclusion in the WHO access group is warranted [1,2,3,4,5]. Where susceptibility is proven, co-trimoxazole is a valuable option in management of Gram-negative infections of the respiratory, genitourinary and biliary tracts, or of the abdominopelvic cavity [150]. Considering the potential applicability of sulphonamides/trimethoprim in systemic infection against the increasing background of resistance, blind use for treatment or prophylaxis of UTI no longer appears tenable and would be expected to contribute to further increases in resistance [96,97,98,99].

3.7. Associations Between Macrolide Use and Resistance

Macrolide use was common in most countries (Table 1). In group 1 countries, mean J01FA consumption decreased from 2.18 to 1.607 ddd/1000/day and in group 2 countries from 3.609 to 3.31 ddd/1000/day (Figure 7). Consumption of macrolides (J01FA) was strongly associated with all 12 resistotypes on univariate analysis (Table 2) and this retained significance on multivariate analysis for all E. coli resistotypes and for MRSA though not for any of the K. pneumoniae resistotypes nor for VRE (Table 3). These findings suggest that allocation of macrolides to the WHO watch category is apt [1,2,3,4,5]. Macrolide use has previously been implicated as a risk factor for colonisation and infection with MRSA and the detection of an association between macrolide consumption and MRSA incidence is not unexpected [151,152,153,154]. The relationship observed here between macrolide usage and increased resistance in E. coli is more surprising. Enterobacterales have an intrinsically high degree of resistance to macrolides, thus it is conceivable that they would be subject to minimal selective pressure from these agents [155]. Frequent detection of macrolide specific resistance determinants in Enterobacterales, often linked with other resistance genes on plasmids, implies that this is not the case [155,156]. Some studies have found that macrolide exposure is an independent risk factor for colonisation or infection with ESBL producing or otherwise multidrug resistant GNB [147,148]. Others report that macrolides have one of the lowest risks amongst antimicrobial classes for resistance selection in Gram-negative pathogens and some investigators have questioned their place in the WHO watch category based upon this [40,97]. It is unclear why the correlation between macrolide consumption and various resistotypes was independently significant for E. coli and MRSA but not for K. pneumoniae and VRE. Whatever the true resistance potential of macrolides, as an antimicrobial class with limited, well defined first line indications, they are undoubtedly overused. One of their main applications is in penicillin allergic patients with CAP or SSTI. It is well established that true penicillin allergy albeit potentially life threatening is massively over diagnosed [144]. Even in CAP patients without a documented allergy, macrolides are oftentimes advocated for use alongside a β-lactam with the rationale that they will cover atypical organisms and may reduce mortality via immunomodulatory mechanisms and/or suppression of bacterial virulence factors such as pneumolysin, a pore-forming exotoxin secreted by pneumococci [159,160,161,162]. It has not been established whether the benefits of adding a macrolide to β-lactam apply generally to CAP patients or only to a subset of CAP patients with the most severe disease or for whom an atypical aetiology is strongly suspected or proven; reports in the published literature are conflicting [163,164]. In many locales, macrolide resistance is now common amongst the principal pathogens causing both CAP (pneumococci and M. pneumoniae) and SSTI (S. aureus and β-haemolytic streptococci) [88,165,166].

3.8. Associations Between Lincosamide Use and Resistance

Mean lincosamide (J01FF) consumption remained stable in both group 1 and 2 countries but was more than twice as high in the latter (Figure 8). Usage levels varied across EEA countries (Table 1). Like macrolides, lincosamides were associated with all resistotypes in E. coli though only with CARBR in K. pneumoniae (Table 3). Lincosamides, unlike macrolides, were not associated with MRSA and in fact, a very weak though significant, negative correlation was observed for this pairing and with VRE (Table 3). The principal lincosamide in clinical use, clindamycin, is counted as an access agent in the WHO AWaRe schema [1,2,3,4,5]. Its independent association with E. coli resistance as observed here, along with substantial risk for provoking C. difficile colitis may imply that this allocation is not deserved. The legitimacy of clindamycin as an access agent has been previously questioned in the literature [40]. Note, however, that although the correlations identified between lincosamide use and E. coli resistance did attain independent significance, the effect size was comparatively modest (Table 3).

3.9. Associations Between Aminoglycoside Use and Resistance

Average aminoglycoside (J01G) utilisation decreased from 0.064 to 0.052 and from 0.133 to 0.12 ddd/1000/day in group 1 and 2 countries, respectively (Figure 8). Countries differed widely in aminoglycoside consumption (Table 1). Although aminoglycoside (J01G) consumption correlated positively with all resistotypes on univariate analysis (Table 2), the significance of this was lost in multivariate analysis, for all but 3 resistotypes namely AGR and 3XR in K. pneumoniae and AGR alone in E. coli (Table 3). It would therefore seem that aminoglycosides, though selecting for resistance to themselves do not appreciably select for resistance to 3GC or FQ in these pathogens, even though the converse occurs with FQ and 2GC/3GC which appear to select for resistance to aminoglycosides as well as to themselves and to each other [167,168,169,170]. Moreover, cross resistance between individual aminoglycosides, depending on the mechanism involved, is often absent in aerobic GNB whilst the same cannot be said for quinolones or cephalosporins [167,168,169,170]. Isolates with resistance to gentamicin, for instance, often remain fully sensitive to amikacin depending on the underlying mechanism [170]. EARS-NET classifies isolates as aminoglycoside resistant if they exhibit resistance to any one of gentamicin, tobramycin or amikacin [73]. Prior to the advent of ultra-broad-spectrum β-lactams and fluoroquinolones in the 1980s, aminoglycosides had been the preferred ‘workhorse’ agents for severe infections due to aerobic GNB but fell out of favour owing to toxicity concerns and the tedious, costly requirement for therapeutic drug monitoring [171,172]. The findings presented here indicate that aminoglycosides do have a comparatively low resistance potential and suggest that the assignment of gentamicin and amikacin to the WHO access group is justified [1,2,3,4,5]. In some areas of the UK and particularly in Scotland, gentamicin has been used first line, with or without amoxicillin, for empirical treatment of undifferentiated sepsis for over a decade, without compelling evidence of increasing resistance or clinical inferiority [137,138]. Empirical use of an optimally dosed aminoglycoside to provide aerobic Gram-negative coverage for up to 96 hours whilst sensitivities are awaited may spare the need for empirical use of broad-spectrum agents in urinary, biliary and intraabdominal sepsis [137,138]. Where ongoing Gram-negative coverage is required beyond this time and use of further doses is thought to risk toxicity, a targeted agent with low resistance potential can be preferentially chosen based on laboratory results. Use of aminoglycosides was twice as high in group 2 countries as in group 1 countries and changed little over 5 years (Figure 8). This might relate to their use as alternatives to carbapenems or polymyxins in multiresistant Gram-negative infections though this is purely speculative. Some carbapenem resistant GNB remain sensitive to one or more aminoglycosides. For instance, KPC producing K. pneumoniae belonging to the prominent ST-258 clone typically retain gentamicin susceptibility [173].

3.10. Associations Between Quinolone Use and Resistance

In group 1 countries, mean quinolone (J01M) use was 3-fold lower at baseline relative to group 2 countries (0.994 vs 3.06 ddd/1000/day) and further declined by approximately one third over 5 years (Figure 7). Two of seven group 2 countries, Bulgaria and Cyprus, increased J01M use by 0.824 and 0.928 ddd/1000/day, respectively, from already high baselines of 2.996 and 5.751ddd/1000/day (Table 1) with the overall effect that mean quinolone consumption in group 2 countries was unchanged over 5 years (Figure 7). In univariate analysis, quinolones (J01M) were strongly associated with all resistotypes (Table 2) and these associations retained independent significance in multivariate analysis for all resistotypes except 3GCR in K. pneumoniae, VRE and unexpectedly, MRSA (Table 3). These findings further validate the view that quinolones are major drivers of resistance and deserve their allocation to the WHO watch group [1,2,3,4,5]. Although having an undoubtedly high resistance potential, the advantages of potent bactericidality, Gram-negative coverage and uniquely high oral bioavailability may tempt clinicians to overuse this class of drugs [174]. The remarkably high resistance potential of quinolones as a class may therefore have to be weighed against the advantages of oral therapy which may include avoidance of the need for hospitalisation and vascular access which are both risk factors in and of themselves for acquisition of drug-resistant nosocomial infections [175]. Whilst it certainly seems from these findings that quinolones as a class have a high resistance potential and do merit inclusion in the WHO watch category, this data does not permit comparisons of resistance potential between individual quinolones. In vitro studies would suggest that the genetic barrier towards de novo mutational resistance is lower for ciprofloxacin than for moxifloxacin in Gram-positive organisms whereas the converse holds true for Gram-negative organisms including Enterobacterales and P. aeruginosa [176]. Some have argued that levofloxacin has a more balanced spectrum and superior pharmacokinetic profile in this regard with an overall lower resistance potential and lower risk for C. difficile colitis though clear evidence for this in clinical practice is lacking [6,7,70].

3.11. Associations Between Glycopeptide Use and Resistance

Mean glycopeptide (J01XA) use increased from 0.057 to 0.067 ddd/1000/day in group 2 countries and stabilised around 0.03 ddd/1000/day in group 1 countries (Figure 8). Consumption of glycopeptides (J01XA) had positive correlations of varying strength and significance for all resistotypes on univariate analysis, however, none of these remained significant in multivariate analysis, including, surprisingly, their association with VRE (Table 2 and Table 3). Literature is conflicting on whether glycopeptide exposure is a risk factor for acquisition of resistant Enterobacterales or VRE [6,7,70]. A limitation of this study is that data was not stratified by route of glycopeptide administration. It is conceivable that orally administered glycopeptides may exert greater selective pressure on the gut flora than their IV counterparts, given that the latter do not appreciably concentrate in the bowel lumen [183,184]. Indeed, it is for this very reason that vancomycin administered orally but not intravenously, is effective in colitis due to C. difficile or staphylococci [184,185]. Conversely, it may also stand to reason that IV glycopeptides would exert greater selective pressure on organisms in other anatomic compartments.

3.12. Associations Between Nitroimidazole Use and Resistance

Group 1 countries had a slight drop in nitroimidazole (J01XD) use from 0.048 to 0.041 ddd/1000/day as did group 2 countries from 0.111 to 0.093 ddd/1000/day (Figure 8). Nitroimidazole (J01XD) use was associated with all resistotypes on univariate analysis, but this retained significance only for VRE and for all 5 E. coli resistotypes on multivariate analysis (Table 2 and Table 3). Lack of association between nitroimidazole consumption and MRSA on multivariate analysis, may owe to the fact that S. aureus, unlike E. coli or Enterococcus spp. does not reside primarily in the gut alongside a predominantly anaerobic microflora vulnerable to disruption by nitroimidazole exposure [69,186,187]. The association of lincosamides, another class of antianaerobic agent, with E. coli resistotypes but not MRSA might be similarly explained (Table 3). Note, however, that lincosamides, were not associated with VRE unlike nitroimidazoles. Nitroimidazoles were more strongly associated with VRE than any other class of agent, including glycopeptides and cephalosporins (Table 3). Intriguingly, the Iberian peninsula reported zero usage of nitroimidazoles (Table 1) and very low rates of VRE (Figure 3) despite having both high consumption of other antimicrobial classes (Table 1) and a high prevalence of other resistant organisms (Figure 1, Figure 2 and Figure 3). Many studies have previously identified metronidazole use as a risk factor for colonisation of with VRE and MDR-GNB [61,62,63,64,65,66,67,68,69]. Strong, independent, associations between nitroimidazole use might suggest that metronidazole is misplaced in the WHO access group [1,2,3,4,5,40]. It should be noted, however, that all the main antianaerobic agents including nitroimidazoles, penicillin/β-lactamase inhibitor combinations and lincosamides were found to entail relatively high resistance risk (Table 2 and Table 3). Why lincosamides and nitroimidazoles were associated with resistance in E. coli but not in K. pneumoniae is unclear given that both are enteric organisms (Table 3). It is possible that this may be explained by the fact that whilst both organisms reside predominantly in the gut, K. pneumoniae does so less exclusively and infections due to this organism may result comparatively more often from environmental sources or contaminated fomites as opposed to autoinoculation or translocation from the host’s own gut [188,189,190,191,192]. K. pneumoniae also occurs proportionately more often in nosocomial infection relative to E. coli and it is possible that shortcomings in infection control procedure contribute more to its spread with lesser influence from antibiotic consumption [190,191,192]. Furthermore, these distinct species may occupy subtly different niches when colonising a host and differ in their interactions, whether competitive or cooperative, with other commensals making up the chiefly anaerobic microbiome of the alimentary canal [193,194,195]. Given that nitroimidazoles are usually used in conjunction with other agents to provide aerobic coverage, this cannot be excluded as a confounding factor. Hoffman and colleagues recently found that in patients undergoing colorectal surgery, prophylaxis with a combination of cefuroxime and metronidazole promoted intestinal carriage of coliforms resistant to carbapenems and/or 3rd generation cephalosporins more so than did prophylaxis with ertapenem alone [196]. Regardless, antianaerobic agents are widely overused and should be an easy target for antimicrobial stewardship initiatives [197]. Most oropharyngeal anaerobes are adequately covered by penicillin G or V alone and routine addition of metronidazole to therapy in peritonsillar abscess or dental infections confers no additional benefit providing adequate drainage is achieved [198,199,200]. In aspiration pneumonia, addition of metronidazole to a penicillin does not lead to better outcomes yet is still widespread practice [201,202]. Except in cases complicated by anaerobic bacteraemia or biloenteric anastomoses, anaerobic coverage is not routinely required in biliary tract infection [203]. Crucially, use of metronidazole is redundant where penicillin/β-lactamase inhibitor combinations, carbapenems, chloramphenicol or tigecycline are used as these agents all offer broad anaerobic coverage [203,204,205,206,207,208]. Aside from the use of adjunctive clindamycin in necrotising SSTI to suppress exotoxin production and downregulate virulence factors in β-haemolytic streptococci, S. aureus and histotoxic clostridia, double anaerobic coverage is almost never clinically indicated and has repeatedly been linked to increased harms without added benefit [204,205,209,210,211,212,213,214].

3.13. Associations Between Nitrofuran Use and Resistance

Average nitrofuran consumption was similar in group 1 and 2 countries and increased slightly in both over 5 years (Figure 6). Individual nations ranged greatly in levels of nitrofuran usage with no use recorded at all for Bulgaria or Slovakia and >4 ddd/1000/day recorded at the other extreme, in Poland (Table 1). Consumption of nitrofurans (J01XE) had no significant associations, positive or negative, with any resistotype, on univariate analysis (Table 2). Some associations between nitrofuran use and certain resistotypes although weak, gained significance in multivariate analysis (Table 3). Specifically, nitrofuran consumption had weak negative correlations with FQR for E. coli and CARBR for K. pneumoniae whilst having weak positive correlations with AMPR in E. coli, with all K. pneumoniae resistotypes other than CARBR and with VRE (Table 3). Nitrofurantoin, though usually effective against E. coli has much less consistent activity against Klebsiella spp. and other enterobacterial genera which may cause urinary tract infection (UTI), especially amongst patients who are catheterised, have calculi or other structural abnormalities of the genitourinary system [215,216]. It appears logical that nitrofurantoin use may shift the aetiology of UTI in favour of these organisms, particularly in the case of hospitalised patients with urological risk factors. Increased reliance on nitrofurantoin as one of exceedingly few options for treatment of UTI due to VRE may account for the positive relationship observed between this resistotype and nitrofuran consumption [217,218]. It could be argued that since nitrofurans are indicated solely for uncomplicated lower UTI and are not useful in serious systemic infections that there is less at stake from resistance towards them than to other agents and that they should thus be used preferentially for this niche application. Nitrofurantoin use has previously been shown in some studies to be inversely correlated with resistance amongst E. coli to other antibiotics commonly used for UTI such as trimethoprim [97]. Nitrofurantoin does not adversely impact the gut flora, presumably because it concentrates exclusively in urine [218,219]. On balance, allocation of nitrofurantoin to the WHO access group seems appropriate [1,2,3,4,5]. Preferential use of nitrofurantoin for treatment of uncomplicated lower UTI is justified though complex lower UTIs with risk factors for involvement of GNB other than E. coli may require alternative therapy and this should be guided by susceptibility testing whenever possible.

4. Conclusions

Analysis of ESAC-NET and EARS-NET data indicates that there are indeed strong associations with overall antimicrobial consumption and the prevalence of key resistance phenotypes in sentinel pathogens, varying in spatiotemporal distribution. Heavy consumption of certain agents namely 2nd and 3rd generation cephalosporins, fluoroquinolones, penicillin/β-lactamase inhibitor combinations, carbapenems, macrolides and nitroimidazoles are a major driver of resistance in EEA countries. Given that much of this use may be for UTI, CAP and uncomplicated SSTI, in which narrow spectrum agents with lower resistance potential would be better suited, this highlights a clear target for antimicrobial stewardship initiatives. Lack of availability, or prohibitive prices, may limit access in certain EEA countries to low resistance potential agents including both β-lactamase labile and stable penicillins (represented by ATCC codes J01CE and J01CF), first generation cephalosporins (ATCC J01DB) and nitrofurans (ATCC J01XE). Addressing inequitable access to such agents may contribute to improvements in antimicrobial stewardship with consequent reductions in antimicrobial resistance. It is demonstrated here that aminopenicillins, though having a generally low resistance potential, are associated with resistance towards amoxicillin/ampicillin in E. coli. Though this may not seem a significant problem given that resistance to these agents has now been widespread for decades, it should be noted that use of these antibiotics is extremely common as are E. coli UTIs. A substantive decrease in use of these agents may therefore yield a significant decline in selective pressure for resistance in E. coli and possibly also in other pathogens innately sensitive to these drugs such as Proteus spp. and H. influenzae. Wider use of narrower spectrum penicillins G & V as alternatives to aminopenicillins, as is practice in Scandinavian countries, merits serious consideration. All agents with broad antianaerobic activity had a high resistance potential with the implication that clinicians and antimicrobial policymakers should carefully consider where anaerobic coverage is needed to avoid unnecessary use of such agents singly or worse still, in redundant combinations. A key strength of this study is its use of a large sample size extracted from data that is freely available in the public domain with power calculated a priori. Though countries outside the EEA were not considered here, it is probable that many of the findings presented could be extrapolated more widely. Whilst one would not wish to stifle an already lacklustre antimicrobial pipeline, it can be noted that many agents in clinical development at present, belong to already known antimicrobial classes with overall high resistance potential such as the quinolones, cephalosporins and carbapenems [220]. The impetus for preferentially investing in such agents exists because their broad-spectrum of activity and wide applicability is convenient for clinicians. This could indicate a repeating historical precedent whereby wider spectrum agents are favoured over narrower spectrum agents. During the last major flurry of antimicrobial development in the 1970s - 1980s, for example, fluoroquinolones and 3rd generation cephalosporins were readily embraced by medics whilst narrower spectrum agents such as temocillin and mecillinam, effective solely against Enterobacterales, and cefsulodin, exclusively targeting pseudomonads, were commercial failures [113,114,118,221,222]. It would seem prudent going forward to closely monitor new agents for resistance potential as they are introduced and to evaluate their microbiotoxicity [223]. Alternative approaches including bacteriophage therapy, bacteriocins, antivirulence compounds, antibodies and immunomodulators have all shown promise in treating infection with potentially less ecological footprint than conventional small molecule antibiotics though rapid and accurate diagnostics may be necessary to facilitate clinical application of many of these agents given the high selectivity of their actions [224,225,226,227]. One key weakness of this study is that consumption of individual compounds was not considered (ATCC level 5) thus differences in resistance potential between different agents belonging to the same class have not been resolved. Factors such as population density, infection control measures, the role of the COVID-19 pandemic in addition to the use of antimicrobials in veterinary medicine and agriculture were not included in this model and the possibility that they have confounded the results to some extent cannot be discounted. A further limitation is that certain lesser used antimicrobial classes were not considered at all, examples would include the polymyxins, amphenicols, streptogramins, oxazolidinones, monobactams, phosphonics and fusidanes [74]. It is quite possible that the resistance potential of an agent could change over time in accordance with evolving microbial aetiologies and resistance patterns. As an illustration of this, fluoroquinolone use is now thought to pose a high risk for acquisition of C. difficile with almost universal consensus [38,228]. This was not always the case, and it is now believed that acquisition of fluoroquinolone resistance by key ribotypes with enhanced virulence, such as 027, changed the epidemiologic situation from one where quinolones were associated with a comparatively minimal risk to the higher risk that is widely acknowledged today [38,228]. Similarly, tetracyclines are now generally thought to carry a minimal risk of selecting MRSA. This was not always so and owes to the fact that currently circulating strains happen to be tetracycline sensitive. The predominant MRSA clones causing problems in the 1960s and 1970s were resistant to tetracycline and rampant overuse of these drugs at that time was thought to have contributed to their success [229,230]. These limitations present opportunities for future studies delineating risk factors for the spread of multidrug resistant organisms.