Our study provides comprehensive information on the utilisation of systemic antibiotics among children and adolescents in the age group ≤18 years in Denmark, Italy, Germany, the Netherlands and the UK during the years 2005 to 2008. Our findings illustrate striking variations of total systemic antibiotic use in paediatric outpatient care between these countries. Substantial differences of outpatient antibiotic use among children across Europe have been described before, but these previous studies only provided comparable data of drug use for up to three countries and suffered from different definitions of drug utilisation measures. Furthermore, comprehensive data about age-group-specific distributions of antibiotic subgroups was lacking for most countries of this study and a comparison has not been conducted so far. In 2001, the European Surveillance of Antimicrobial Consumption Project was established to gather reliable and comparable information on the utilisation of antibiotics in Europe, however, without distinguishing between adults and children . The current study captured outpatient systemic paediatric antibiotic use of five countries in different European regions ensuring high inter-country comparability, due to consistent definition of drug utilisation measures, age groups and classification of antibiotic subgroups.
Overall, the annual antibiotic prescription rates in the Italian region Emilia Romagna were more than three times higher than those in the Netherlands, the country with the lowest prescription rates, and still substantially higher than those in Germany, the country with the second highest use. When compared to other studies, magnitude of paediatric antibiotic use in Italy exceeded use reported for Canada (608 prescriptions per 1000 children <15 years of age in 2003)  and Sweden (764 prescriptions per 1000 children 0-6 years of age in 2002)  as well, but appears to be comparable to the U.S. (910 prescriptions per 1000 person years in children <18 years of age in 2001) .
High antibiotic prescribing in the Italian outpatient setting compared to the other countries in our study might be related to differences with regard to historical backgrounds, cultural and social factors, awareness about antibiotic resistance in the community and among healthcare providers  as well as the ability of physicians to adequately diagnose common infectious diseases . So far, reasons for strong variations of antibiotic use across European countries have not yet been fully investigated. Nevertheless, previous studies suggest that awareness about antibiotic resistance  and inadequacy of antibiotics to treat viral infections  is poor among Italian patients and perception of parent expectations by Italian physicians is a major determinant of antibiotic prescribing to children .
In contrast, several previous studies showed antibiotic utilisation in the Netherlands to be lowest in Europe, overall and in the paediatric setting [6, 13]. The Netherlands are a country with a strict prescribing policy for antiinfectives, and there are intensive efforts into promoting guideline-appropriate prescribing habits to combat antibiotic resistance .
Although antibiotic use was by far the highest among Italian children and adolescents, antibiotic prescription rates in Denmark, Germany and UK still exceeded those in the Netherlands to a great extent. These observed strong variations of total paediatric antibiotic use among the countries of study are unlikely to reflect an actual therapeutic need which would have to be based on marked differences in the burden of infectious diseases between these countries. This assumption is also supported by the observed pronounced increases of prescription rates during winter months which were expectedly highest in Italy, and smallest in the Netherlands. Increases of antibiotic use are most likely related to seasonal rise of predominantly viral respiratory infections and hence should be limited .
Since our findings could not provide information beyond the 4-year study period, we compared our prescription rates with those of other studies which included other study years or longer time periods. In this respect, our data for the years 2005-2008 in the Netherlands agreed well with the findings by de Jong et al. who reported a variation of the total annual number of antibiotic prescriptions between 282 and 307 per 1,000 Dutch children in the years 1999-2005 . This suggests an overall stable total antibiotic use among Dutch children for almost ten years. Gagliotti et al. observed annual prescription rates per 1,000 person years among children 0-14 years of age from Emilia Romagna, varying between 1,158 and 1,358 during 2000-2002 . This is in line with our findings in children below 15 years of age of 1,123 (2007) and 1,034 prescriptions per 1,000 person years (2008), indicating marginal changes over time of total paediatric use in Emilia Romagna. Prescription rates among British children did not show any apparent trend towards lower or higher prescribing in our study over the study years. Gradual annual increases of prescription rates between 2000 and 2007 were reported in the UK based on data from the General Practice Research Database (GPRD) . However, differences to our findings for the years 2005 to 2007 were small and might have resulted from variations in the regional distribution of general practices contributing data to THIN and/or the GPRD. We observed a steady decrease in prescription rates in Germany during 2005-2008. Another German study also based on GePaRD data found slightly higher prescription rates among German children without an obvious downward trend for the years 2004-2006 . This former study, however, included data from four rather than three health insurances, resulting in a study population of about twice as many children as in this study which may explain the difference.
We also detected remarkable differences in the choice of antibiotic subgroups between the countries of our study. Narrow spectrum penicillins formed the majority of systemic antibiotics in Denmark, whereas prescriptions of broad spectrum penicillins were most frequent in the four other countries. In line with that, the highest agent-specific prescription rates were reported for phenoxymethylpenicillin in Denmark, amoxicillin in Germany, the Netherlands and the UK and amoxicillin plus enzyme inhibitor in Italy. Relatively high use of narrow spectrum penicillins in Denmark in comparison to other European countries has also been reported previously . However, it is noteworthy that even though proportions for narrow spectrum penicillins were highest in Denmark, broad spectrum penicillins formed the antibiotic subgroup most frequently prescribed to children in the age group ≤4 years in all 5 countries. This might be due to frequent use of amoxicillin or amoxicillin and enzyme inhibitor in the treatment of acute otitis media, which shows the highest incidence in the first two years of life .
Macrolides were commonly prescribed in all five countries with the highest use in the age groups 10-14 and 15-18 years. Relative proportions of macrolide use were lowest in Denmark. This finding is in agreement with a Danish practice guideline which recommends restricting the use of macrolides to patients with penicillin allergies in the treatment of common childhood infections . Several studies from the U.S. and Europe show a strong association of high macrolide use and the emergence of resistant strains of pneumococci and other common pathogens [23–25]. Hence high prescription rates of macrolides are questionable and likely to unnecessarily increase selective pressure on bacterial pathogens. In particular high use of clarithromycin and azithromycin in the Emilia Romagna region appears unjustified, since international guidelines do not recommended these agents as first-line treatment of common childhood infections [29–32]. Furthermore, longer plasma half-life of azithromycin and clarithromycin in contrast to erythromycin might even accelerate the emergence of antibiotic resistance [33, 34].
Our findings regarding paediatric cephalosporin use are in line with previous studies which reported strong variations of cephalosporin prescribing across Europe, with the lowest prescription rates in the Netherlands and Denmark [13, 18–20, 35]. Overall, the prescription rate of cefaclor (a second generation cephalosporin) in German children was the second highest after amoxicillin, and use of second generation cephalosporins was particularly common in very young children. Only in Italy, the parenterally administered third generation agent ceftriaxone was prescribed frequently. Considerably higher prescribing of parenteral antibiotics in Italian outpatient care in contrast to Northern European countries has been reported previously . The high relative use of cephalosprines in Germany and Italy as observed here, suggests frequent prescribing of these antibiotics as a first–line treatment of common paediatric respiratory infections. This is in conflict with international practice guidelines [29, 30] recommending that cephalosporins should be preserved for second-line treatment in cases such as treatment failure of first-line agents, non-type 1 allergy to penicillins or unusually severe symptoms.
Strengths and limitations
Our study overcomes limitations of previous studies and facilitates the comparison of paediatric antibiotic prescriptions in five countries based on a common protocol using the same drug utilisation measures. It provides insight into the age-group-specific distributions of antibiotic subgroups in the paediatric setting of the participating countries. Ascertainment of antibiotics prescribed in the outpatient setting was complete in all databases except Denmark, where some antibiotics as e.g. cephalosporins are reimbursable only in particular circumstances and might therefore have been underascertained. Nevertheless, given that the Danish National Health System reimburses antibiotics for the entire spectrum of childhood indications,  the proportion of antibiotics which could not be captured due to private prescribing appears to be small. Besides this, differences of antibiotic use across countries reflect differences in prescribing behaviour of outpatient providers and not in the type of data.
Our study has some limitations, which have to be taken into consideration. First, for this study only data for the years 2007 and 2008 was available from the Northern Italian region Emilia Romagna. Hence, insight into the development of antibiotic prescribing over time is limited. However, our findings are in good agreement with Gagliotti et al.  In addition, extrapolation from our findings to Italy in general is not straight forward, given considerable regional differences of prescribing patterns in Italy. Nonetheless, previous studies about marked heterogeneity of antibiotic use across Italy with up to 19% higher paediatric prevalence rates of antibiotic exposure in southern regions compared to Emilia Romagna  indicate, that overall paediatric antibiotic use in the Italian outpatient setting during the years of our study might have been even higher than suggested by our findings.
Since all five databases only provide information on drugs prescribed in the outpatient setting, antibiotics administered to inpatients to treat severe childhood infections could not be studied. Given that indications underlying the issued prescriptions were not available in all databases, the appropriateness of single treatment courses could not be assessed. Additionally, compliance with the antibiotic prescription remains unknown.