Quantity of Research Available
A total of 352 citations were identified in the literature search. Following screening of titles and abstracts, 307 citations were excluded and 45 potentially relevant reports from the electronic search were retrieved for full-text review. No potentially relevant publications were retrieved from the grey literature search. Of these potentially relevant articles, 40 publications were excluded for various reasons, while five publications met the inclusion criteria and were included in this report. Appendix 1 describes the PRISMA flowchart of the study selection.
Additional references of potential interest that did not meet the selection criteria are provided in Appendix 5.
Summary of Study Characteristics
Five publications were identified for inclusion in this report: two systematic reviews with meta-analyses,7, 8 two randomized controlled trials (RCTs),9, 10 and one non-randomized study.11
A detailed summary of study characteristics is provided in Appendix 2.
Study Design
The systematic review by Shah et al.7 included six primary studies published between 2007 and 2013. Four of the six studies with unreported study designs provided data to be pooled in the meta-analysis; two studies had a quasi-experimental design and were excluded from the meta-analysis because the relative risks for the study populations could not be calculated. The literature search for the meta-analysis by O’Horo et al.8 captured studies published up until May 23, 2011. A total of 12 studies conducted between 2005 and 2010 were included in the review, including one RCT and 11 studies with a quasi-experimental design.
The open-label RCT by Boonyasiri et al.9 used block randomization to assign patients in groups of four to either the intervention or control groups. The study by Noto et al.10 was a cluster-randomized, controlled crossover study. Each of the five adult intensive care units (ICUs) was randomized at the start of the study to either the intervention or control treatment for 10 weeks followed by a two-week washout period and a crossover to the other treatment; each ICU switched three times, for a total of four 10-week periods of alternating intervention and control treatments.
The non-randomized study by Chung et al.11 had an interrupted time series design, in which the outcomes were measured monthly over the 13 month control period and then again over the 11 month intervention period.
Country of Origin
The two meta-analyses7, 8 and one RCT10 were conducted by authors in the United States of America, one RCT9 was conducted in Thailand, and the non-randomized study11 was conducted in South Korea.
Patient Population
All five publications that met inclusion criteria for this review recruited adult patients admitted to critical care settings. The setting for one study11 was limited to medical ICUs, and the remaining four studies recruited patients from multiple units, including:
Coronary or cardiac ICUs
7–10Respiratory care units
7, 9Bone marrow transplantation units
7, 8
One systematic review8 considered ICU and non-ICU settings; of the 12 included studies, 11 were conducted in various ICUs, and one study was conducted in long-term care acute hospitals. The study conducted in the long-term care acute hospitals evaluated a different preparation of CHG and was not included in the review’s subgroup analysis of 2% CHG-impregnated washcloths.
Interventions and Comparators
All studies evaluated bathing with 2% CHG-impregnated washcloths; three studies examined commercially prepared 2% aqueous CHG-impregnated cloths manufactured by Sage Products8, 10 or GAMA Healthcare,11 one systematic review did not specify the source of the CHG wipes,7 and one study used 2% CHG-impregnated wipes prepared by the study hospital’s pharmacy department but not otherwise described.9 When reported, CHG bathing was performed daily.8–11 One meta-analysis8 included several CHG formulations, but reported a subgroup analysis of 2% CHG-impregnated washcloths (evaluated in seven of 12 studies included in the meta-analysis). Comparators included daily bathing with soap and water,7, 8 disposable non-antimicrobial washcloths,7, 8, 10 non-medicated wet towels,11 or standard of care (not otherwise defined).8 One study compared daily CHG bathing with twice daily bathing with soap and water.9
Outcomes
The identified studies evaluated the incidence and prevalence rates of several health care-associated infections, such as: bloodstream infections (BSIs),8 central-line associated bloodstream infections (CLABSIs),7, 9, 10 ventilator-associated pneumonia (VAP),9, 10 catheter-associated urinary tract infections (CAUTIs).9, 10 and Clostridium difficile (C. difficile) infections.10 Other outcomes of interest were related to transmission of MDR bacteria, including MDR bacteria-related favourable events (i.e., eradication of bacterial infection or persistently negative bacterial swabs throughout the ICU stay),9 and rates of carbapenem-resistant Acinetobacter baumannii (CRAB).11 Some studies also reported length of stay in the hospital and ICU9, 10 and adverse skin reactions.8, 9, 11
Summary of Critical Appraisal
Strengths of both systematic reviews7, 8 were related to the comprehensive literature search and duplicate study selection with a reported consensus mechanism. Duplicate data extraction was also performed in one meta-analysis,8 but it was unclear whether this was done in the review by Shah et al.7 In addition, both meta-analyses used appropriate statistical methods to pool results, and reported on the statistical heterogeneity of included studies. However, the review by Shah et al.7 suffered from unclear reporting; patient characteristics of the four studies in the meta-analysis were insufficiently described, making it difficult to assess the clinical similarities or differences, and the appropriateness of pooling data from these studies. The reporting of the study characteristics of the two studies excluded from the meta-analysis and narratively summarized were also minimal. Furthermore, while the authors stated that six studies qualified for inclusion in the critical appraisal, the quality assessments of these included studies were not provided, limiting confidence in the review conclusions based on the study findings. In contrast, the systematic review and meta-analysis by O’Horo et al.8 described the characteristics of the individual studies in sufficient detail, and the conclusions acknowledged the limitations of the findings due to their derivation from non-randomized studies. Additional strengths of this review were the statistical and graphical assessment of publication bias, and the provision of funding information and conflict of interest declarations for the individual studies and the review authors.8 Limitations common to both meta-analyses were the lack of an a priori published research objectives or a protocol, as well as the absence of an excluded studies list.7, 8
The two RCTs9, 10 and the non-randomized study11 were generally reported well, with clearly described study objectives, inclusion and exclusion criteria, interventions, comparators, and main outcomes. Two of the three studies were randomized,9, 10 and one RCT9 concealed the random sequence by using sealed, opaque envelopes. The second RCT10 used random number generation to assign treatment groups by unit at the beginning of the study; however, it was not indicated whether this assignment strategy was concealed from the staff at the time of random number generation. The participants were not blinded to the intervention in any of the studies, but the outcome assessors were blinded in the two RCTs.9, 10 It was unclear whether outcome assessors in the non-randomized study11 were unaware of the treatment groups; blinding those analyzing data is feasible and would reduce the chance of bias in the direction of the intervention. All three studies were conducted in settings and in conditions that would be representative of usual care for the majority of patients in an ICU, but two studies9, 11 excluded patients who were admitted to the ICU for less than 48 hours. While this is reflective of the intent to focus on new, hospital-acquired infections, this may exclude some patients who would still require bathing and who may represent patients with significantly different characteristics than those admitted to an ICU for longer periods of time. Appropriate statistical tests were used in all studies, although there was limited description of patient characteristics in the study by Chung et al.,11 leaving it unclear whether major confounders were considered in the data analysis. The main findings were clearly reported in two studies;10, 11 however, one RCT9 presented the infection rate without reporting simple outcome data (e.g., total number of catheter-days per group) or estimates of the variability in the data, which limits interpretation of the results. In addition, patients who died within 48 hours, were missing culture data, or those who transferred to another ward were excluded from the analysis in this study.9 These patients are unlikely to be representative of those remaining in the trial, and it is unclear whether their losses to follow-up were related to the study treatments. Finally, a power calculation was performed for the two RCTs9, 10 to determine the number of study patients required to provide at least an 80% chance of detecting a statistically significant difference between groups.
Additional details regarding study strengths and limitations are provided in Appendix 3.
Summary of Findings
Five publications7–11 regarding the clinical effectiveness of CHG wipes for infection prevention in adult patients in a critical care setting met the inclusion criteria for this review. No relevant studies were identified regarding the clinical effectiveness of CHG wipes for infection prevention in adult patients in acute care, or regarding the cost-effectiveness of the use of CHG wipes in acute or critical care settings; therefore, no summary can be provided for these questions.
What is the clinical effectiveness of chlorhexidine gluconate wipes for infection prevention in adults patients in critical care?
The studies included in this review reported results pertaining to rates of health care-associated infections,7–10 MDR bacterial transmission,9, 11 length of stay in the hospital and ICU,9, 10 and adverse events.8–11
Additional details regarding study findings are provided in Appendix 4.
Health care-associated infections
Four of the identified studies reported on the acquisition of health care-associated infections, including any BSI8 and CLABSIs in particular,7, 9 VAP,9 and CAUTI.9 To account for the potentially rare nature of individual health care-associated infections, one study reported a composite primary outcome including CLABSI, CAUTI, VAP, or C. difficile infection.10 In this study, the expected number of ICU admissions over the study period (approximately 10, 000 patients) would provide at least 95% power to detect a difference in the primary outcome of 0.1 infections per 1,000 patient-days.
The results from two meta-analyses showed that the relative risk of CLABSIs or BSIs (including but not limited to CLABSIs) in patients bathed with 2% CHG-impregnated washcloths was 0.417 and 0.468 times that of the control group, respectively; the control groups in these studies represented a mixed selection of comparators including soap and water, non-antimicrobial washcloths, and standard of care (not otherwise described). One of the reviews excluded two studies from the meta-analysis but summarized the results narratively; these study findings supported the meta-analysis results, suggesting that CHG bathing was associated with a decrease in CLABSIs in ICU patients.7 However, a recent RCT that compared daily bathing of ICU patients with 2% CHG-impregnated washcloths with twice daily bathing with soap and water found no significant difference in the incidence of CLABSIs between the two treatment groups.9 Likewise, this study found that there was no difference between groups in the incidence of VAP and CAUTIs.9
The RCT that reported a composite primary outcome across all ICUs (including the surgical unit) also found no significant difference in the composite infection rate between CHG and disposable non-antimicrobial washcloths.10 This result was consistent when individual infection rates were compared between CHG and control groups for the combined ICU population. Furthermore, a subgroup analysis of the primary outcome by ICU confirmed that there was no significant difference in infection rates between treatment groups in any individual unit.
MDR bacterial transmission
In one RCT,9 the proportion of patients with MDR bacteria-related favourable events appeared to decrease from day 3 to day 14 in both groups (44% to 35% in the control group, 36% to 29% in the CHG group), but there was no significant difference in the number of favourable events between treatment groups at any time point. This study also showed that the overall prevalence of carbapenem-resistant enterobacteriaceae was not significantly different between the CHG (4.8%) and control (4.0%) groups at any time point up to day 14, or at any swab site.
The interrupted time series study11 evaluated the rates of CRAB before and after the implementation of daily bathing with 2% CHG-impregnated washcloths in a medical ICU, and found a significant decrease in the CRAB prevalence and incidence density in the CHG bathing period compared with the control period. However, after the initial decreases from the end of the control period to the beginning of the CHG period, there was no further reduction in the CRAB prevalence or incidence density levels or trends over the CHG period (11 months).
Length of stay
Two RCTs reported that daily bathing with 2% CHG-impregnated washcloths had no significant impact on length of stay in either the ICU or the hospital, when compared with a control bathing method.9, 10
Adverse events
Three studies reported on adverse skin reaction events.8, 9, 11 One systematic review and meta-analysis8 with an unclear total number of included patients reported that adverse event data was available in six studies. Among the seven studies included in the subgroup analysis of 2% CHG-impregnated washcloths, a total of six patients from three studies had rashes, one of which was possibly related to CHG.8 Adverse event data were not available for the control groups in these studies. One RCT indicated that 2.5% of patients in the CHG group experienced mild skin reactions, while skin reactions were not reported for the control group.9 No adverse skin reactions were reported during the CHG bathing period of the interrupted time series study.11
Among all secondary outcomes analyzed by ICU in the RCT by Noto et al.,10 the unadjusted in-hospital mortality rate in the trauma ICU was the only outcome that was significantly reduced in the CHG group compared with the control group. However, this difference between treatment groups was eliminated after adjustment for the expected mortality rate.
Limitations
The primary limitation of this review is the lack of identified evidence regarding the clinical effectiveness of CHG wipes in a non-ICU acute care setting, and regarding the cost-effectiveness of CHG wipes in acute or critical care. Furthermore, few primary studies were identified that were conducted in non-surgical settings to be included in this report; both systematic reviews included some studies from surgical ICUs, which may be notable if these patients are anticipated to respond differently to the use of CHG wipes than other ICU patients.
One of the three primary studies11 and the majority of studies included in the two systematic reviews7, 8 identified for this report were non-randomized before-and-after studies, in which there may be differences beyond the intervention, such as variation in staff practices or patient populations, that could have contributed to the observed differences between the CHG and control groups in these studies. Data from the two RCTs9, 10 that randomized patients to CHG or control bathing within the same period of time may be more reliable, and these studies did not observe differences infection rates or MDR bacteria-related events between treatment groups. However, results from these RCTs may also be subject to influence by the structured nature of clinical trials that may differ from real-world clinical practice, or potentially altered staff or patient behaviour in light of study observation. The study by Noto et al.10 was described as a pragmatic effectiveness trial rather than an efficacy trial, meaning that the study treatments were performed as part of routine patient care rather than by research personnel according to a strict study protocol, thereby reducing the risk that results may have been unrealistic due to study design. However, as a result of this pragmatic approach, compliance with the intervention was not assessed and the authors acknowledged that the potential impact of this factor was unknown. In addition, the RCT by Boonyasiri et al.9 compared once daily CHG bathing with twice daily bathing with soap and water; it is unclear whether the lack of observed difference between groups was due to the similar clinical effectiveness of the two methods, the potential benefit of increased bathing frequency, both, or another unknown reason. Therefore, results from all studies should be interpreted with caution.
The most recent systematic review by Shah et al.7 excluded the RCT by Noto et al.10 from the meta-analysis because it was judged not to meet inclusion criteria regarding study methodology; specifically, the rates of health care-associated infection were low to begin with in that hospital and may not be generalizable to other ICUs. However, Noto et al.10 evaluated a composite primary outcome to account for rare individual infection events and performed a power calculation to increase the likelihood of observing a true difference between groups, suggesting an attempt to address this limitation by study design. It is therefore unclear whether the exclusion of this RCT from the meta-analysis by Shah et al.7 was justified, which is particularly important given the conflicting results reported by these two studies.