Antibiotic use and resistance in public-sector hospitals in KwaZulu-Natal.

OBJECTIVE To investigate a possible association between level of care, antibiotic use and antimicrobial resistance in 16 public-sector hospitals providing different levels of care in KwaZulu-Natal. DESIGN A multicentre surveillance study was undertaken in 16 hospitals at three progressive levels of health care (district, regional, tertiary) where each hospital submitted 100 consecutive, non-repetitive isolates judged in the laboratory to be of potential clinical significance. Isolates were identified and susceptibility testing was undertaken using the Kirby-Bauer disc diffusion method with minimum inhibitory concentrations (MICs) extrapolated on an automated reading system. Isolates were grouped according to their natural resistance profiles, and percentage susceptibility, mean percentage susceptibility and standard deviation to each antibiotic were stratified within and across hospital levels. Antibiotic use data were expressed as the number of daily divided doses (DDDs) per 1000 patient-days. SETTING Two tertiary, 9 regional and 5 district public hospitals in KwaZulu-Natal. OUTCOME MEASURES Percentage susceptibility. RESULTS There was a general trend among the 1 270 isolates of highest susceptibility in district hospitals, followed by regional and then tertiary hospitals. This is consistent with the referral system where health conditions become increasingly severe/complex requiring greater antibiotic use and broader-spectrum agents at progressive hospital levels, with statistical significance (p < 0.05) evident where sample numbers were relatively large. Trend variations could be associated with the qualitative and quantitative differences in antibiotic use, albeit without statistical corroboration. Three per cent of the total number of isolates were sensitive to all antibiotics tested and 6% were resistant to a single agent only. The remaining 91% showed acquired resistance to more than one drug. The standard deviation ranged from 0% to 55%. CONCLUSIONS This study showed that resistance profiles among bacteria varied greatly within and across hospital levels. While antibiotic use varied as much, a statistically significant correlation between use and resistance could not be established. It was therefore postulated that the effect of selection pressure was obscured by other resistance determinants apparent in public hospitals in resource-poor settings. On a clinical level, the study showed that resistance profiles among bacteria vary too much to allow a national antibiotic policy as proposed in the standard treatment guidelines. Rather, such guidelines should be directed to specific profiles found in different hospitals and at different levels of health care. Regular surveillance to adjust such guidelines is essential.