Study objectives: Multiple guidelines exist for recommending initial antibiotic choice in the treatment of outpatient community-acquired pneumonia (CAP). Broad-spectrum macrolides, as well as doxycycline, are most often cited as appropriate first-line drugs. The guidelines also list fluoroquinolones as a possible initial antibiotic agent, but all discourage their use in otherwise healthy patients because of concerns about fluoroquinolone resistance. We determined the trends in antibiotic choice for treatment of CAP in the emergency department (ED) setting at a large academic medical center.

Methods: A retrospective 3-year medical record review of consecutive patients discharged to home from the ED with a diagnosis of CAP was performed. CAP patients were included during the peak pneumonia season for 3 consecutive periods. Period 1 was October 1, 2000, to March 31, 2001; period 2 was October 1, 2001, to March 31, 2002; and period 3 was October 1, 2002, to March 31, 2003. A standardized form was completed by trained medical students and antibiotic choice was documented. Using the mean proportions for the 3 periods, pairwise comparisons between 2 periods for the 2 antibiotics (macrolides and fluoroquinolones) were performed using the Bonferroni adjustment to adjust for multiplicity. An interaction effect, antibiotic versus time, was included in the linear regression model. In addition, data were obtained from the microbiology laboratory concerning percentage of drug-resistant Streptococcus pneumoniae. The proportions of susceptibility for 4 years (2000 to 2003) were compared. Pairwise comparisons were performed between 2 years for susceptibility using the Bonferroni adjustment to adjust for multiplicity.

Results: The proportion of macrolide use significantly decreased (79% to 71% to 59%), and the proportion of fluoroquinolone use significantly increased (7% to 16% to 21%) during the 3 periods. The comparisons reached statistical significance when period 1 was compared with period 3. Macrolides and fluoroquinolones (in percentage use) are negatively correlated with Pearson's correlation coefficient (–.6981; P<.01). The interaction effects between the percentage use of macrolides and fluoroquinolones versus time were significant. The regression coefficient for the interaction effect between macrolides and time was negative, whereas the regression coefficient for the interaction between fluoroquinolones and time was positive. The pairwise comparisons for the 4 years of susceptibility data showed that there were not significant changes in the prevalence of drug-resistant S pneumonaie.

Conclusion: Within a 3-year period, fluoroquinolone use at this academic institution is increasing, whereas macrolide use is decreasing, a trend not in accordance with published guidelines. This trend is concerning considering the emergence of fluoroquinolone-resistant S pneumoniae.