Annual macrolide prescription rates and the emergence of macrolide resistance among Streptococcus pneumoniae in Canada from 1995 to 2005

Over the last 20 years, Canadian pneumococcal surveillance studies have documented a steady rise in macrolide resistance. In the current study, we probed the nature of associations between the emergence of macrolide-resistant Streptococcus pneumoniae in Canada and changes in macrolide (azithromycin, clarithromycin and erythromycin) prescription rates. Macrolide susceptibility testing data for respiratory tract isolates of S. pneumoniae (n = 15 109) were acquired from two published national Canadian surveillance databases, and dispensed outpatient macrolide prescription data were acquired from the proprietary Intercontinental Medical Statistics (IMS) Health Canada CompuScript database. Nationally, macrolide resistance increased from 3.7% in 1995 to 19.0% in 2005 (P = 0.003) as the annual macrolide prescription rate increased from 106.7 to 123.2 prescriptions/1000 persons per year (P = 0.003). From 1995 to 2005, azithromycin and clarithromycin prescriptions increased from 4.8 to 52.5 prescriptions/1000 persons per year (P < 0.0001) and from 24.7 to 58.4 prescriptions/1000 persons per year (P = 0.005), respectively, whilst erythromycin prescriptions decreased from 77.2 to 12.3 prescriptions/1000 persons per year (P < 0.0001). By univariate regression analysis, increasing rates of azithromycin (R2 = 0.931; P < 0.0001) and clarithromycin (R2 = 0.725; P = 0.0009) prescriptions and a decreasing rate of erythromycin prescriptions (R2 = −0.963; P < 0.0001) were all associated with increasing macrolide resistance from 1995 to 2005. Multivariate regression analysis showed that a model including all three macrolide prescription rates provided the best fit to the trend of increasing macrolide resistance. When the data were analysed by provincial origin, no statistically significant associations were found between prescription rates of any macrolide and macrolide resistance rates by univariate and multivariate regression analyses. We conclude that increasing macrolide resistance among respiratory isolates of pneumococci in Canada from 1995 to 2005 was associated both with decreasing prescriptions for erythromycin and concurrent increases in prescriptions for azithromycin and clarithromycin (azithromycin > clarithromycin by univariate regression analysis). Resistance development is complex and factors other than macrolide use may also be associated with observed increases in macrolide resistance in Canada from 1995 to 2005.