Sources of systematic errors in the epidemiology of vancomycin-resistant enterococci

Abstract

The rates of vancomycin resistance among enterococci (VRE) have been increasing worldwide. However, reports on vancomycin-resistant enterococci (VRE) are easily biased and meta-data reporting is insufficient. Additionally, no standardised protocol for VRE testing currently exists. The aim of our study was to investigate, for the first time, the impact of introduced bias in VRE reports. We also analysed the sensitivity of our in-house screening test for VRE, namely, a broth-enriched PCR assay. Retrospective review of microbial and clinical data on all patients tested for VRE who had been admitted to a large university hospital over a 5-year period and an analysis of the possible impact of introduced bias. Our screening test was also evaluated using clinical isolates. A total of 27,636 screening tests were carried out over the 5-year study period, of which 2,459 were VRE-positive. The number of screening tests increased dramatically over the study period, with 1,053 tests carried out on 435 patients in 2006 and 9,444 tests carried out on 5,104 patients in 2010. VRE prevalence was 8.1 % over the 5-year period. The introduction of measurement bias caused a clear overestimation of absolute VRE numbers. The sensitivity of our screening test was 95.5 % with a positive predictive value of 39 %. Biased reports lead to the implementation of high-cost containment measures that may be both unnecessary and detrimental to the patient. Our data show that systematic errors in VRE reports caused a clear overestimation of absolute VRE numbers, thereby indicating an outbreak situation even though the actual prevalence of VRE was decreasing. We suggest that reports of VRE must take measurement and analysis biases into account, otherwise any conclusion drawn is unreliable and inconclusive.