ORGANIZATION: The University HealthSystem Consortium (UHC; www.uhc.edu) hospitals and Virginia Commonwealth University. PROBLEM OR ISSUE ADDRESSED: Nationwide rates of bacterial resistance among nosocomial pathogens are increasing. Mathematical models and limited clinical data suggest that greater diversity (or heterogeneity) of antibiotic use may decrease rates of resistance. GOALS: The goal of this investigation is to determine if an increase in diversity of antibiotic use at UHC hospitals will be associated with an improvement in rates of resistance. The UHC is an alliance of 97 academic medical centers and affiliated hospitals. UHC hospitals that also subscribe to the Clinical Resource Manager program were asked to participate in a long-term, multihospital antimicrobial use and surveillance program in 2003. We retrieved claims data for systemic antibiotic use administered to adult inpatients beginning in year 2002 (N=31 hospitals) and through 2005 (N=35 hospitals). Using 5 classes of antibiotics that are commonly administered for treatment of nosocomial infections (aminoglycosides, 3/4th generation cephalosporins, fluoroquinolones, carbapenems, and b-lactamase inhibitor combinations) we applied Simpson’s Index of Diversity (D) to arrive at a diversity measure for each hospital. We also collected annual antibiograms from most hospitals for 2002 through 2004, and we focused on methicillin-resistant Staphylococcus aureus (MRSA) and fluoroquinolone-resistant Pseudomonas aeruginosa (FQRPa). We determined both the proportion of resistant isolates (% resistant) and the rate of resistance isolates (no. organisms/1000 admissions) for MRSA and FQRPa. Hospitals with a Diversity Index less than the median diversity value for year 2004 (i.e., D <0.722) were offered an educational intervention. The intervention consisted of a visit to each hospital by the principal investigator (PI, REP) at the investigator’s expense. During the visit, a 45 minute educational presentation was given to the hospital antibiotic committee regarding the potential merits of diversity of antibiotic use, and to provide benchmarking information regarding diversity and resistance in MRSA and FQRPa to each hospital. OUTCOMES ITEMS USED IN THE DECISION: The hypotheses to be tested and their outcomes included: (1) Would hospitals extend an offer for the PI to visit and give a presentation, (2) in response to the intervention, would hospitals purposefully change antibiotic use toward greater diversity, and (3) would resistance rates and proportions of resistant MRSA and FQRPa improve following a change toward greater diversity of use. IMPLEMENTATION STRATEGY: An e-mail was sent in the summer of 2006 to pharmacy and infection control personnel at 17 hospitals with low diversity (above). This text of the message included a brief review of the potential value of diversity, each hospital’s diversity score and an offer to provide the intervention (above). The intervention had to occur by the end of 2006, and 5 hospitals responded positively. The last of 5 interventions was made on 12/5/06. RESULTS: It is too early to say if hospitals will adopt policies to change antibiotic use that will result in greater diversity of use during 2006, but it appears doubtful for the following reasons: (1) The “diversity hypothesis” does not seem to be familiar to many of the contact persons, (2) the potential benefit of diversity of antibiotic use is not yet established; only two small-scale prospective investigations have been published, (3) antibiotic formularies and antibiotic use appear to be driven more by acquisition costs of therapy than by concerns regarding resistance and (4) programs to implement greater diversity will probably require an increase in the use of relatively expensive antibiotics, such as the carbapenems. LESSONS LEARNED: (1) In retrospect, it may have been preferable to select hospitals for intervention based on resistance rates (e.g., resistance rates greater that the median value), rather than on diversity of antibiotic use (i.e., D less than the median value). This strategy may have generated more offers for the intervention. (2) The costs of bacterial resistance are not well studied and an argument that a reduction in rates of resistance will carry cost benefits is difficult to make at this time. (3) The methods that can be used by a hospital that will result in greater diversity of antibacterial drug use and that are relatively straight-forward and practical are not yet worked out. Computer guided antimicrobial selection will be the most feasible method to implement greater diversity, but this is not yet commonly available. In spite of these limitations we will continue to monitor antibiotic use and rates of resistance through 2009 in order to examine relationships between diversity of use and changes in rates of bacterial resistance. In addition we will monitor and measure other relevant and potentially confounding variables such as infection control efforts, hospital policy toward antibiotic use (e.g., formulary restriction, etc.) and the impact of community acquired (CA) infections on rates of nosocomial resistance, such as CA-MRSA.