Correlates of antibiotic use in Taiwan hospitals

OBJECTIVE: To determine factors that correlate with increased antibiotic use among adult inpatients in Taiwan. DESIGN: Retrospective survey of medical records. SETTING: 14 acute-care hospitals (8 regional hospitals, 6 medical centers) in Taiwan. PARTICIPANTS: A systematic probability sample from each hospital, totaling 663 adult inpatients who were discharged or had died in early 1999. MEASUREMENTS: Infectious disease physicians at the 14 hospitals collected data from medical records regarding patient demographics, hospitalization, discharge diagnosis, and antibiotics received. RESULTS: A total of 447 (67%) patients received antibiotics for an overall rate of 813 antibiotic-days (number of days patients received each antibiotic)/1,000 patient-days. Both the proportion of beds in intensive care units ([ICUs] Pearson correlation coefficient [r], 0.67; 95% confidence interval [CI95], 0.36-0.89; P<.01) and the proportion of patients admitted to surgical services (r, 0.66; CI95, 0.20-0.88; P=.01) correlated with the mean patient rate of antibiotic-days/hospital-day (MPAUD). In contrast, we found no correlation between the proportion of patients who received antibiotics and the MPAUD. Using multiple linear regression, medical center status was the only independent predictor for increased MPAUD (regression coefficient [b], 0.15; CI95, 0.05-0.24; P<.01). There was no correlation between pooled rates of antibiotic-days/hospital-day and any hospital demographic factors. First-generation cephalosporin (39%) and aminoglycoside (24%) use accounted for the majority of antibiotic-days. CONCLUSIONS: Antibiotic use is greater in medical centers than in regional hospitals and appears to be independent of surgical case mix or the proportion of ICU beds. Determination of multiple, independent measures of antibiotic use may be necessary to understand the relation between antibiotic use and resistance in hospitals. ? 2001 The Society for Healthcare Epidemiology of America.