Analyzing antibiotic consumption data is a difficult task. Although it is agreed that unified classification systems and measurements such as DDDs are needed for international comparison, there are still unresolved issues concerning how to measure change over time within one country accurately and how to retrieve the values closest to reality. By looking only at consumption expressed in DDD, change in the consumption patterns could be over- as well as underestimated if the average grams per package change over time. The use of packages as a proxy for prescriptions seems to be appropriate to avoid this bias.
Despite all technical difficulties, surveillance of antimicrobial use is a key strategy to monitor appropriateness of antimicrobial therapy. In Austria, consumption is relatively low (12.3 – 14.6 DID) in comparison to other European countries (mean DID 19.04) and the USA (24.9 DID) [7]. In Europe, the level of consumption ranges from 9.8 DID in The Netherlands to 33.4 in Greece. In the past ten years, the level of antibiotic consumption has not changed substantially, regardless of the measurement used. Compared to other European countries, Austria ranges among the low consumer countries such as The Netherlands, Germany and Estonia for penicillins as well as for total consumption [8, 9]. With regard to cephalosporins and fluoroquinolones, Austria ranges mid-field in Europe, meaning a relatively high proportion of fluoroquinolones and cephalosporin compared to other European countries [10, 11]. Although the total volume of consumption has not changed over time, the composition of substances did. There was a considerable increase in the consumption of fluoroquinolones, betalactams and a decrease in the use of tetracyclines and sulfonamide/trimethoprim. The level of consumption of macrolides did not change over time.
Analysis of the consumption of particular substances shows that especially the use of ciprofloxacin and amoxicillin/clavulanic acid has increased dramatically, indicating that substances with a lower selective pressure, such as penicillin, have been exchanged for amoxicillin/clavulanic acid [8]. The percentage of use of moxifloxacin has increased continuously, from 15% in 2004 to 20% of total fluoroquinolone use in 2007 [11]. In 2001, the patent for ciprofloxacin expired and since then, the number of prescriptions has more than doubled. Interestingly, the increase of ciprofloxacin prescriptions is almost equal to the magnitude of the decrease in prescriptions of norfloxacin and sulfonamide/trimethoprim together, indicating a switch in the therapy of urinary tract infections in primary care. The increased use of fluoroquinolones and the traditionally frequent use of third-generation cephalosporines [10, 12] are very likely to cause and promote the incidence of resistant microorganisms in Austria [13]. It is remarkable that the resistance rate of invasive E. coli to fluoroquinolones rose from 7% in 2001 to 25.5% in 2007; resistance to third-generation cephalosporins also rose from 0% in 2000 to 7% in 2007. Ena et al. showed that the use of fluoroquinolones and urinary catheterisation are independent risk factors for the acquisition of urinary tract infections with ESBL-producing E. coli [14]. This and other evidence in the literature suggest a close association between the use of fluoroquinolones and the increase in the incidence of multidrug-resistant E. coli [15, 16].
Routinely-collected surveillance data on antibiotic resistance usually do not include those from uncomplicated urinary tract infections. Due to the lack of representative surveillance data from urinary samples, data on invasive infections must be used for the development of empirical therapy regimens, although resistance might be overestimated, since 32% of E. coli bacteraemia cases are health-care associated or even hospital-acquired. With regard to complicated urinary tract infections, there is evidence from the US that first-line therapy with sulfonamide/trimethoprim might be inadequate due to resistance rates of up to 24%; therefore, it is recommended to use ciprofloxacin instead [17]. In Austria, the rate of sulfonamide/trimethoprim resistance is known only from local data. In our hospital for example, the resistance rate in urinary samples was 23.3% for sulfonamide/trimethoprim and 15% for fluoroquinolones in 2006. Fluoroquinolone resistance in invasive E. coli isolates is high, with up to one third of the isolates being resistant; therefore, fluoroquinolones can not be recommended for first-line therapy. Furthermore, a study from Italy shows that resistance to fluoroquinolones in E. coli has a negative impact on the outcome of community-acquired urinary tract infections (UTI) and their use in this indication should therefore be avoided [18].
S.pneumoniae remains the most important pathogen for community-acquired pneumoniae [19]. The susceptibility results of invasive pneumococci indicate that penicillin G is still very effective, since there is no upwards trend and the number of isolates with high-level resistance is very low (5/323). On the other hand, the macrolide resistance rates do not show a very favourable situation: 15% of all invasive pneumococci isolates were resistant to macrolides. Compared to Germany, where up to 28% of prescribed substances are narrow-spectrum penicillins [8], in Austria, this substance group accounts for only 7% of prescriptions. According to Garcia-Suarez et al., resistance rates in invasive S. pneumoniae are lower than resistance rates in non-invasive isolates, which is why these data should be interpreted with care [20].
Limitations of the study
Although the increase in fluoroquinolone consumption and the increase of fluoroquinolone resistance coincide, which suggests an association, the results should be interpreted with care. For one, this is an ecological study that considers only aggregated country data for consumption and resistance. Another reason is that it is not possible to make a distinction between hospital and community-acquired infections, which might lead to overestimation of resistance rates. Using only resistance data from hospitals and not taking into account resistance data from the community might add to any overestimation of resistance rates. Furthermore, the lack of consumption data from the hospital setting neglects the possible influence of hospital prescribing on the evolution of resistance.