The estimated average coverage of the populations of countries participating in EARSS is 52% (range 10% to 90%) (13). Laboratories that participate in EARSS screen invasive S. pneumoniae isolates for oxacillin resistance (14). When an isolate is found to be nonsusceptible, the EARSS protocol requests confirmation as intermediate- or high-level resistance to penicillin by determination of MICs. Laboratories perform microbiologic testing and interpret results according to their own standards. National guidelines in Europe differ; isolates of S. pneumoniae are considered nonsusceptible to penicillin if the MIC is >0.06 (15–18) or >0.12 (19,20) mg/L. For this report, we use nonsusceptibility and intermediate resistance as synonyms; PNSP isolates are either intermediate or fully resistant to penicillin. Only the first invasive isolate per patient per quarter is reported.

To assess the comparability of susceptibility test results, a quality assurance exercise was performed in September 2000 among 482 laboratories from 23 countries participating in EARSS. The concordance (agreement of reported results with intended results) for the detection of penicillin resistance in the three S. pneumoniae control strains was 91% (21). Laboratories sent standardized data to the national EARSS data manager, who checks data contents and ensures conformity with the EARSS data format. In collaboration with WHO, an export module from the laboratory-based software WHONET was developed for EARSS (22). Every quarter, data are forwarded to the central database at the National Institute of Public Health and the Environment (RIVM), Bilthoven, Netherlands, where the project is coordinated.

National outpatient sales data for antibiotics from 1997 were purchased from IMS Health Global Services, London, United Kingdom, for 13 of the 15 member states of the European Union. Corresponding data were obtained from the Danish Medicines Agency for Denmark and from the National Corporation of Swedish Pharmacies for Sweden (23). The IMS data were examined and adjusted according to the Anatomic Therapeutic Classification (ATC) system used by WHO (24). The amount in kilograms for an antimicrobial agent was converted to a number of defined daily doses (DDD). The DDD, which is based on the average daily dose used for the main indication of the drug, is appropriate for comparisons of drug use over time and in different geographic areas. For beta-lactam antibiotics, we combined ATC groups J01C (extended- and narrow-spectrum penicillins) and J01D (cephalosporins); macrolides were classified under code J01F. No data were available for the combination of trimethoprim and sulphonamide.

We considered nonadherence of patients to the physician’s prescription in individual countries as a possible confounder of antimicrobial resistance. Branthwaite et al. reported nonadherence levels from a population-based survey in seven countries (25). Data from four of the seven countries (Spain, Belgium, the United Kingdom, and Italy) were also captured in EARSS.

We calculated the proportion of PNSP among all invasive S. pneumoniae isolates from each country reported during 1998-99. Because probabilities allow only values between 0 and 1, we modeled the natural logarithm of the odds of PNSP resistance (logodds).

Least-square linear regression analysis was used to assess correlation between antimicrobial use (of beta-lactam antibiotics and macrolides, expressed in DDD per 1,000 population per day) and the logodds of resistance. We correlated nonadherence levels with the logodds of resistance in the same way. We calculated the Spearman coefficient of determination (r-square) and its corresponding p value. For the calculation of the regression lines, we weighted the data points by the inverse of the variance of each data point. We used SAS software (SAS Institute Inc., Release 6.03., Cary, NC).

During 1998-99, 337 laboratories from 11 European Union member states (Belgium, Finland, Germany, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, and United Kingdom) and one nonmember state (Iceland) reported 4,872 invasive S. pneumoniae isolates to EARSS. The proportion of PNSP among isolates of invasive S. pneumoniae ranged from 1% to 34% (Table) (Figure 1). Southern European countries reported higher rates than northern European countries.

Data on outpatient sales of beta-lactam antibiotics and macrolides were available for 1997 from all 15 European Union member countries. Antimicrobial use varied widely between countries. Sales to outpatients ranged from 3.8 to 23.6 DDD per 1,000 inhabitants per day for beta-lactam antibiotics and from 0.97 to 5.98 DDD for macrolides. The three countries with the highest reported use were France, Spain, and Portugal for beta-lactam antibiotics and France, Spain, and Italy for macrolides; the three countries with the lowest use were the Netherlands, Germany, and Austria for beta-lactam antibiotics and Sweden, the Netherlands, and Finland for macrolides.

For 11 countries, information was available for both antimicrobial resistance and antimicrobial use. Linear regression of the correlation of use of beta-lactam antibiotics and the logodds of resistance showed an r-square of 0.80 (p=0.0002) (Figure 2). The equation for the regression isFigure 2

The logodds of resistance to penicillin among invasive isolates of Streptoccus pneumoniae (PNSP; ln(R/[1-R])) is regressed against outpatient sales of beta-lactam antibiotics in 11 European countries; antimicrobial resistance data are from 1998 to 1999 and antibiotic sales data are from 1997. DDD = defined daily dose; BE = Belgium; DE = Germany; FI = Finland; IE = Ireland; IT = Italy; LU = Luxembourg; NL = the Netherlands; PT = Portugal; ES = Spain; Se = Sweden; UK = United Kingdom.