We conducted a hospital-wide point prevalence survey to measure the prevalence of VRE, as our previous surveillance activities were targeted to patients felt to be at risk and therefore may have resulted in selection biases. In this survey, we confirmed that vanB VRE is endemic in inpatients at our hospital. We found a lower prevalence in short stay areas, such as the emergency department (3%) and the short stay unit, but prevalence was 8-29% in other inpatient areas. This is higher than reported by previous studies conducted at Melbourne hospitals[11, 17–19]. This is also in contrast to previous studies demonstrating that VRE colonisation was restricted to particular high risk inpatient areas. Our finding that 80% of the VRE colonised patients in the survey were newly detected cases demonstrates that the previous screening strategy of surveying patients only on wards when a new clinical isolate of VRE was detected is inadequate.
Prima facie, our results suggest that antibiotic selection pressures appear to play a larger role in determining VRE colonisation than cross-transmission at our institution. Exposure to antibiotics, particularly meropenem, was strongly associated with VRE colonisation; this is consistent with previous studies suggesting that antibiotic regimens with activity against anaerobic bacteria are potent risk factors in the development of VRE[20–22]. The vanB gene is known to be present in commensal enteric bacteria such as Clostridium, and transmission of this gene to E. faecium has been demonstrated in vivo. In other studies, a variety of other antibiotics have been implicated as risk factors for VRE colonisation, including vancomycin, metronidazole, piperacillin-tazobactam, ticarcillin-clavulanate and third generation cephalosporins[24–26]. However, our analysis comparing meropenem and non-meropenem antibiotic use with no antibiotic use showed the exposure to antibiotics other than meropenem is also associated with VRE colonisation. Thus, our study may have been underpowered to detect differences at the level of individual antibiotics other than meropenem.
We found ambiguous evidence regarding the extent of cross transmission of VRE, which has been shown to be important in previous studies[27, 28]. We did not find proximity to other VRE colonised patients to be a risk factor for VRE detection. However, this does not exclude the possibility of cross transmission contributing to endemicity, as transmission may have occurred prior to the time of the survey and would not be detected with frequent patient movements. Conversely, the finding that ribotyping only demonstrated two large clones of VRE in patients does not necessarily implicate extensive cross transmission. If transmission was significant, we would expect a lower diversity among individual wards; however, we found that in all wards with at least two VRE colonised patients had at least two ribotypes. We did not have access to pulsed field gel electrophoresis, regarded as the gold standard for epidemiological investigations of VRE.
The purpose of this study was to examine “downstream” exposures that had implications for VRE control policies, rather than patient-specific “upstream” risk factors that are not modifiable. However, we did find older age and longer length of stay in hospital as independent predictor of VRE colonisation. This is reflected in the low colonisation rate in patients with a short length of stay, including emergency and short stay units.
A strength of this study is that we conducted a hospital-wide survey including a broad range of patients, thus minimizing selection biases. A limitation of this cross-sectional study is that we were not able to determine the timing of acquisition of VRE in the patients identified as being colonised; therefore, patients may have acquired VRE in the remote past, with antibiotic use merely amplifying existing colonisation. In addition, we do not have data on patient co-morbidities, previous invasive procedures, and indwelling devices, which may have confounded our findings. We chose to perform rectal swabs due to practical reasons but previous studies have demonstrated that the sensitivity of these specimens is around 79%, we might have missed detection of few patients with low VRE density[31, 32]. We did not perform separate broth enrichment before the agar plate culture, which may lead to an underestimation of the true prevalence. This may have resulted in misclassification bias in assigning VRE colonised patients as controls and bias the case control study towards the null hypothesis. As the sensitivity of VRE detection is likely to be related to bacterial density, it is likely that detection of VRE in screening is the product of two processes, that of new acquisition of VRE (whether by endogenous generation or exogenous exposure) and amplification of existing low level colonisation by antibiotics.