Physicians Tend to Use National Rather Than Local MRSA Infection Data for Antibiotic Prescribing

Alice Goodman

September 17, 2010

September 17, 2010 (Boston, Massachusetts) — American physicians appear to be using national surveillance data on bacterial infections rather than local data to guide their prescribing for methicillin-resistant Staphylococcus aureus (MRSA) wound infections, according to a study presented here at the 50th Interscience Conference on Antimicrobial Agents and Chemotherapy.

"Physicians are relying on national data more than on state data to prescribe antibiotics for wound infections when making an educated guess before getting culture results. During this window of opportunity, state level data are a better guide. An educated guess should be informed by state and, if possible, local data," asserted lead author of the study, Marin L. Schweizer, PhD, epidemiologist and associate professor at the Carver College of Medicine at the University of Iowa in Iowa City.

Her study was based on 2 national databases: the Surveillance Network, an electronic database of antibiotic susceptibilities from 133 zip codes covering the years 1999 to 2007; and the IMS Health's Xponent database, a prescription tracking database created by IMS Health that covers more than 70% of all prescriptions in the United States, including clindamycin, linezolid, and trimethoprim-sulfamethoxazole (TMP). IMS Health provides global market information to pharmaceutical and healthcare industries.

Dr. Schweizer and coauthors assessed 107,215 MRSA-positive wound cultures and 106,641,604 antibiotic prescriptions. They correlated the rate of MRSA wound infections with zip code, state, and national data, and correlated that with antibiotic prescribing according to zip code, state, and national data.

No correlation between the number of infections and prescribing patterns was observed at the zip code level, but she said that was to be expected because of the "noise" at that level, meaning that 0 patients could have MRSA in one month, 3 in the next month, 0 in the third month, and so on.

Some correlation was observed at the state level between positive wound cultures and prescribing patterns for clindamycin and linezolid, but not for TMP. The strongest correlation was seen between national MRSA rates and prescribing of clindamycin and linezolid. "These are the most common outpatient antibiotics prescribed," Dr. Schweizer said.

"Relying on national data can lead to inappropriate prescribing. State and local data should guide prescribing. For example, if there are 0 cases of MRSA in one zip code area or in the state during a particular month, then the physician should consider other pathogens," she explained. "Giving the correct antibiotic early in the course of infection would save money and improve outcomes."

Dr. Schweizer and colleagues are planning to conduct a needs assessment to determine what kinds of data are most useful to physicians for prescribing antibiotics correctly, and then to determine methods of getting them the appropriate data.

Local Data Important for Appropriate Prescribing

"For empiric therapy, the physician's choice of initial therapy should be guided in part by local flora and resistance patterns," stated David Weber, MD, professor of medicine, pediatrics, and epidemiology at the University of North Carolina, Chapel Hill.

The University of North Carolina hospital provides a yearly antibiogram, which lists organisms and susceptibilities, he said. These data are subdivided according to community-acquired infections, infections on the hospital floor, and intensive care unit infections. Not all hospitals have the resources to do this, and under those circumstances, the physician should rely on data generated by the state, Dr. Weber said.

Dr. Schweizer has disclosed no relevant financial relationships. Dr. Weber reports receiving support from the Policy Analysis Inc and Forest Research Institute.

50th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC): Abstract K335. Presented September 13, 2010.