DOI: 10.1097/CCM.0b013e3181a5e48e
,
,
PMID: 19448452
Issn Print: 0090-3493
Publication Date: 2009/06/01
Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Excerpt
Compliance to this simple, fast, and effective procedure remains low among healthcare workers. Other measures that may prevent the nosocomial transmission include improved antibiotic stewardship, staff cohorting, maintenance of appropriate staffing ratios, reductions in length of hospital stays, contact isolation, active microbiological surveillance, and better staff education. Currently, the efficacy of many of these individual infection-control interventions remains speculative. Many studies reporting improvement in infection control outcomes are “before–after” studies or involve simultaneous institution of several of these measures, making it impossible to tease out the effects of any of the individual components.
In this issue of Critical Care Medicine, Climo et al (1) examined the association between daily chlorhexidine bathing and the incidence of colonization and bloodstream infections due to MRSA and VRE among ICU patients. They found that colonization was significantly lower when chlorhexidine rather than nonantiseptic soap bathing was used.
The scientific rationale is persuasive: chlorhexidine kills bacteria, whereas soap and water do not. The authors appropriately used complex statistical analyses including time series, Cox models, and Poisson regressions, and they acknowledged potential limitations in their analyses and subsequent conclusions, including the potential for chlorhexidine resistance. Although two previous studies (2, 3) conducted by the Bob Weinstein group at the Cook County Hospital had already pioneered the source control effect of chlorhexidine bathing (i.e., to decontaminate patients’ skin), this study clearly increases the level of evidence. First, the study was conducted by other investigators, outside the Chicago area, which provides a proof of concept regarding feasibility and replication by other, with consistent results. Second, the authors found a strong association between the intervention and the reduction of VRE colonization (−50%, p = 0.008), despite a lower baseline incidence rate (4.35 vs. 26 per 1000 patient-day in the study by Vernon et al [2]). Third, the authors investigated a hard outcome: bacteremia. Similar to the colonization results, they were able to demonstrate significant decreases of VRE bacteremia during the intervention period. The results regarding MRSA bloodstream infection are, however, mitigated by the low baseline incidence.
The main limitation of their study, which applies to infection control research in general, lies in the clustering behavior of VRE or MRSA cases. A cluster refers to a grouping of observations that are related temporally and in proximity. In the context of clinical research, the lack of independence between observations represents a considerable nuisance effect that can bias study results, when the intervention is administered to a group of patients located at the same place, at the same time. The authors did check for autocorrelation and found a Durbin-Watson test value of 1.4 for MRSA acquisition and 2.7 for VRE acquisition. Durbin-Watson value always lies between 0 and 4, and a value of 2 indicates no autocorrelation. On the other hand, only administration of daily chlorhexidine bathing to a group of patients can capture the effect of “colonization pressure” reduction. Consequently, an individually randomized study would not be appropriate in this context, as discussed elsewhere (2).
How could researchers switch from “quasi-experimental” to “experimental” studies? A pure cluster randomized trial may require a lot of clusters to balance clusters (i.e.
