The authors concluded that it is unclear whether restricting vancomycin use is effective in reducing vancomycin-resistant enterococcus colonisation and infection in U.S. hospitals. Although the search was somewhat restricted and more steps could have been taken to minimise the risk of bias during the review process, given the apparent lack of well-designed studies these cautious conclusions appear justified.

To determine the effect of reducing vancomycin use through prescribing interventions on the rates of vancomycin-resistant enterococcus (VRE) colonisation and infection in hospitals in the USA.

PubMed (including MEDLINE), the Cochrane Library and Web of Science were searched to March 2006 for studies published since 1987; the search terms were reported. The reference lists of retrieved articles and reviews were also screened. Inclusion was restricted to published articles in English.

Randomised and quasi-randomised controlled trials were eligible if they reported the prevalence or incidence of VRE colonisation or infection, following any intervention that successfully reduced vancomycin use, compared with rates in control groups not exposed to the intervention. Studies had to be conducted in a U.S. hospital setting and focus on population-level associations.

The participants in the included studies were adults or children, mostly in general in-patient wards or intensive care units, who had been exposed to either endemic VRE of several years’ duration (where stated) and/or an outbreak. The type, timing and sequence of the interventions differed. All studies included restriction of vancomycin (and in some cases other antimicrobials), and some also included infection control measures. The controls were patients in the same ward or hospital, prior to the implementation of the intervention. Vancomycin use was generally measured by the amount of vancomycin prescribed or inappropriately prescribed, and VRE outcomes were reported as the rates of colonisation and/or infection. The studies used various methods of sampling and outcome ascertainment and reported prevalence and incidence in differing ways, including both point prevalence and rates over time. The duration of follow-up ranged from 4 months to 8 years.

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

The authors stated that the validity of each study was assessed using informal methodological criteria, but did not clearly describe how the assessment was performed or how many reviewers were involved.

Pre- and post-intervention data were compared to calculate the relative risk of VRE colonisation or infection at the two time points. The 95% confidence interval and p-value were also calculated where possible. In each study, all reported data for the pre- and post-intervention periods (respectively) were combined to calculate the overall rate for each time period. It was assumed that sample sizes were constant over time, where this was not reported. Where vancomycin use varied over time, only data from study periods or study sites in which vancomycin use was reduced were included.

A single reviewer, who was not blinded, extracted the data.

Clinical and methodological heterogeneity between the studies precluded a meta-analysis, thus the data were combined in a narrative. Study findings were summarised by counting the number of studies reporting a statistically significant increase, decrease or no change in VRE rates, and by noting the range of point estimates in each case. Heterogeneity was explored by stratifying the studies according to the use of cointerventions, study site (ward or hospital), clinical severity of the participants, outbreak status (endemic or outbreak), study design and duration of the intervention.

Thirteen pre-test post-test studies were included: nine had a single pre-test and four had double pre-tests; nine had multiple post-tests.

In all studies the comparison groups differed temporally (i.e. pre- and post-intervention) and few studies compared the characteristics of the two groups. All but two studies were conducted at a single institution and most studies failed to adjust for confounders.

The findings for VRE rates were inconsistent. Seven studies reported statistically significant reductions of 45 to 82.5% in rates of VRE colonisation and/or prevalence and three reported no significant changes. Three studies reported increased rates of VRE of up to 475%, but only one of these reported the statistical significance (p<0.001).

The only statistically significant findings of the stratified analyses were that studies of specific wards, with sicker patients, were more likely to report positive results than studies of whole hospitals which included general in-patients (p=0.029, Fisher’s exact test).

It is unclear whether restricting vancomycin use is effective in reducing VRE colonisation and infection in hospitals in the USA.

The review question and inclusion criteria were clear in most respects, although it was unclear whether there were specific criteria for the vancomycin-reducing intervention. Relevant sources were searched for studies but, as unpublished and non-English studies were excluded, the review is prone to language and publication biases; potential publication bias was not assessed. It was not stated how the study selection or validity assessment processes were conducted and a single reviewer extracted the data, which increases the potential for error and bias. The decision not to meta-analyse the data appears appropriate in view of the heterogeneity between the studies, which was appropriately explored by stratified analyses and was well-addressed in the text. The authors outlined a number of other limitations with the data, notably the poor design of the included studies, with inadequate controls for confounders. Although the search was somewhat restricted and more steps could have been taken to minimise the risk of bias during the review process, given the apparent lack of well-designed studies the authors’ cautious conclusions appear justified.

Practice: The authors did not state any implications for practice.

Research: The authors stated that high-quality studies using experimental designs are needed with vancomycin use as the sole intervention. They suggested that as randomised controlled trials may be ethically unacceptable, a suitable alternative could be the use of segmented regression analysis.

National Institutes of Health, Fogarty International Center, grant number TW006563.

This record is a structured abstract written by CRD reviewers. The original has met a set of quality criteria. Since September 1996 abstracts have been sent to authors for comment. Additional factual information is incorporated into the record. Noted as  [A:....].