Search | Search by Center | Search by Source | Keywords in Title
Parada JP, Conway W, Johnson S, Hur K, Evans CT, Weaver FM, Gerding DN. All Antibiotics Are Not Created Equal - Lessons Not Learned and Unintended Consequences of an Antibiotic Formulary Change. Poster session presented at: Society for Healthcare Epidemiology of America Scientific Annual Meeting; 2009 Mar 20; San Diego, CA.
Background: Clostridium difficile infection (CDI) is a common nosocomial infection closely associated with antibiotic (AB) use. However, ABs do not have equal CDI risks. The Veterans Administration (VA) changed its "workhorse" AB from levofloxacin (levo) to gatifloxacin (gati) in 3/2004 (cost savings), and then from gati to moxifloxacin (moxi) in 2/2006 (gati-related adverse events). We compared CDI events between the levo, gati, and moxi periods. Methods: This is a matched case-control study from 10/2001-9/2006 of 3 VA hospitals which underwent successive simultaneous system-wide AB formulary changes. We compared patients with CDI to those without CDI in their exposure to selected ABs (particularly the fluoroquinolones levo, gati, and moxi), and anti-acid secretion medications (proton pump inhibitors, H2 blockers). Cases and controls were matched 1:1 using propensity score methods. Results: 83,355 patients had an exposure to at least one of the study drugs, corresponding to 35,631 unique acute care patient visits. We identified 1,683 positive C. difficile toxin assays for 1,213 acute inpatient visits. 893 were new or non-recurrent CDI cases corresponding to 822 unique patients with single CDI events. CDI incident rates increased from 9.8/10,000 patient days (95% CI = 8.8-10.9) (levo) to 16.0 (95% CI = 14.3-17.8) (gati), to 18.2 (95% CI = 15.1-21.8) (moxi) (p 0.0001). Figure 1 shows cumulative incident rates for the levo, gati and moxi periods. Table 1, Part A shows the CDI attack rates by study medication, as well as a sub-analysis of CDI risk for patients who received only monotherapy (no other medication prescribed). 808 CDI patients were matched to 808 non-CDI control patients. Compared to controls, CDI cases were more likely to have been exposed gati or moxi than levo (Part B). Conclusions: Our 5-year study involving over 83,000 patients found that CDI rates/incidence increased markedly after the levo to gati formulary change, suggesting that even small differences in an AB's potential to induce CDI are magnified when these AB are used in large quantities. The sustained CDI risk for the moxi period reinforces this point. The matched-control analysis of over 1,600 patients confirmed an increased CDI risk associated with the broader spectrum floroquinolones. A possible confounder is introduction of the epidemic BI/NAP1/027 strain (1st documented in 1/2005). CDI may be one unanticipated consequence of AB formulary changes, and consideration of such indirect AB-associated costs should be factored into planning formulary changes.