Persistence of antimicrobial resistance in respiratory streptococci.

Objectives: To assess whether persistence of antimicrobial resistance (i.e. non-susceptible resistance status) after treatment with penicillins or cephalosporins versus macrolides or tetracyclines differs and to compare the results obtained using routinely collected data with findings reported in prospective studies.

Methods: Routinely collected microbiological data from 14 voluntary participating laboratories (2005) containing information on resistance status and individual antimicrobial consumption patterns (mid 2004-2005) were analysed using a generalised estimating equation (GEE) approach. The link function was adjusted to acknowledge that the proportion of resistant isolates in the population not treated with antibiotics [baseline resistance (BR)] is not necessarily zero. To optimise the comparability of this study with prospective studies, the analysis was repeated after removal of 14 isolates from patients who did not survive 2005.

Results: BR estimates were unstable and their confidence intervals were wide, which called for a sensitivity analysis using an adjusted GEE model with three different BR estimates. All models indicated that the proportion of susceptible isolates differed by treatment group and increased significantly over time, with this increase being independent of treatment group. Persistence of resistance after exposure to macrolides or tetracyclines was approximately three times as long as after exposure to penicillins or cephalosporins.

Conclusions: Resistance following treatment with macrolides or tetracyclines persists longer than following treatment with penicillins or cephalosporins, which confirms the findings from prospective studies and suggests the use of routinely collected data as a valuable alternative to determine such differences in persistence of resistance.