Systematic literature analysis and review of targeted preventive measures to limit healthcare-associated infections by meticillin-resistant Staphylococcus aureus

R Köck1, K Becker2, B Cookson3, J E van Gemert-Pijnen4, S Harbarth5, J Kluytmans6, M Mielke7, G Peters2, R L Skov8, M J Struelens9, E Tacconelli10, W Witte11, A W Friedrich (alex.friedrich@umcg.nl)12 1. Institute of Hygiene, University Hospital Münster, Münster, Germany 2. Institute of Medical Microbiology, University Hospital Münster, Münster, Germany 3. Division of lnfection and lmmunity, University College London, London, United Kingdom 4. Faculty of Behavioural Sciences, University of Twente, Enschede, the Netherlands 5. Infection Control Program, University of Geneva Hospitals and Medical School, Geneva, Switzerland 6. Department of Medical Microbiology and Infection Control, VU University Medical Centre, Amsterdam and Amphia Hospital Molengracht, Breda, the Netherlands 7. Robert Koch Institute, Department for Infectious Diseases, Berlin, Germany 8. Department for Microbiology and Infection Control for Microbiological Surveillance and Research, Statens Serum Institut, Copenhagen, Denmark 9. European Centre for Disease Prevention and Control, Stockholm, Sweden 10. Division of Infectious Diseases, Department of Internal Medicine I, University Hospital Tübingen, Tübingen, Germany 11. Robert Koch Institute, Reference Centre for Staphylococci, Wernigerode, Germany 12. Department of Medical Microbiology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands


Review articles
Systematic literature analysis and review of targeted preventive measures to limit healthcare-associated infections by meticillin-resistant Staphylococcus aureus Meticillin-resistant Staphylococcus aureus (MRSA) is a major cause of healthcare-associated infections in Europe. Many examples have demonstrated that the spread of MRSA within healthcare settings can be reduced by targeted infection control measures. The aim of this systematic literature analysis and review was to summarise the evidence for the use of bacterial cultures for active surveillance the benefit of rapid screening tests, as well as the use of decolonisation therapies and different types of isolation measures. We included 83 studies published between 2000 and 2012. Although the studies reported good evidence supporting the role of active surveillance followed by decolonisation therapy, the effectiveness of singleroom isolation was mostly shown in non-controlled studies, which should inspire further research regarding this issue. Overall, this review highlighted that when planning the implementation of preventive interventions, there is a need to consider the prevalence of MRSA, the incidence of infections, the competing effect of standard control measures (e.g. hand hygiene) and the likelihood of transmission in the respective settings of implementation.

Background
Meticillin-resistant Staphylococcus aureus (MRSA) is a major cause of healthcare-associated infections in Europe. In 2008, the European Centre for Disease Prevention and Control (ECDC) estimated that a total number of 171,200 nosocomial MRSA infections are acquired annually in the Member States of the European Union (EU), and in Iceland and Norway, resulting in 5,400 attributable excess deaths, more than 1 million excess days of hospitalisation and EUR 380 million excess in-hospital costs [1]. The burden of MRSA infections was also shown in an analysis of data on healthcare-associated infections collected prospectively from European intensive care units (ICU) between 2005 and 2008, where 1.7% of all patients developed S. aureus pneumonia or bloodstream infections. A mean of 35% of these infections were caused by MRSA. Moreover, the hazard ratio for mortality was 5.6-times higher (95% confidence interval (CI): 3.4-9.4) for patients with MRSA bloodstream infection than for patients without S. aureus bacteraemia [2].
Among the proposed methods to prevent MRSA, many (e.g. hand hygiene and transmission-based precautions) have been used for general infection control, and their effectiveness has been reviewed extensively [3,4]. However, there is an ongoing discussion about the evidence for the effectiveness of several more specific prevention methods which, nevertheless, have been included in standards for the prevention and control of MRSA in a majority of European countries [5]. Therefore, the scope of this review was to analyse systematically recent literature (published after 2000) with respect to the following questions related to MRSA prevention and control: 1. Does screening of patients before or on admission reduce the incidence of MRSA infection or transmission? How do PCR-based rapid tests for the direct detection of MRSA from screening specimens influence the incidence of MRSA colonisation or infection compared with culture-based methods? 2. Does the decolonisation of nasal MRSA or S. aureus carriage using mupirocin nasal ointment, alone or in combination with other agents, reduce colonisation or the development of infections? 3. Does isolation in single rooms of patients colonised or infected with MRSA prevent the spread of MRSA better than the use of transmission-based precautions (hand hygiene, gloves, aprons) alone? What is the effect of pre-emptive isolation of risk patients for MRSA carriage (until screening results are available)?

Methods
A systematic literature analysis and review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [6]. To identify relevant publications, PubMed, EMBASE and Scopus were searched for articles published between 1 January 2000 and 31 October 2012 in English language. The search terms were: MRSA AND (prevention OR control OR prophylaxis OR preventive measures OR preventive therapy OR preventive treatment OR precaution OR screening OR active surveillance OR decolonization OR mupirocin OR surveillance culture* OR chromogenic OR PCR OR polymerase chain reaction OR rapid test OR isolation OR hygiene OR efficien* OR effective*) AND (healthcare OR hospital OR nursing home OR long-term care facilit*); the search terms were adapted for search in EMBASE: "MRSA AND decolonization", "MRSA AND isolation", "MRSA AND screening".
Titles and abstracts were screened independently by two reviewers (RK and AWF). Studies with outcomes measuring the incidence of MRSA colonisation or infection were included. Exclusion criteria were: Studies that did not report on the effects of the preventive measures on infection or transmission; studies performed in settings other than hospitals, long-term care facilities and nursing homes; case series, outbreak reports and        reviews (the literature lists of the reviews were manually screened for additional relevant publications).
Data were extracted by AWF and RK independently using a standardised form. The study designs were assigned according to a modified study design scheme published by the Centre for Reviews and Dissemination at the University of York, United Kingdom, in the NHS economic evaluation database handbook from 2007. Formal assessment of the quality of studies was not performed. Due to the different study outcomes included, formal meta-analysis was considered inappropriate. Heterogeneity in methodology and outcome measures also prevented quantitative assessment of publication bias.

Results
The literature search identified 9,340 articles, 151 of which were retrieved as full texts after review of titles and abstracts. Of these, 69 articles fulfilled the criteria for inclusion and a further 14 articles were added after search through the literature lists of excluded review articles (Figure). Overall, 83 articles were included in the review .

Screening
We identified 41 studies that investigated the question whether screening for MRSA carriers before or on admission had an impact on MRSA acquisition or infection rates (Table 1)  .

PCR-based screening
Sixteen studies used PCR-based screening techniques in their intervention phases, including one RCT, two prospective cohort studies and 13 comparative studies [7,8,11,13,14,[16][17][18][22][23][24][25][26]29,39,41]. One study reported ambiguous results [16]. Among the studies reporting a decrease of infection or transmission, five compared the intervention group (PCR-based screening) to a control group without active surveillance, with non-compulsory active surveillance or with screening of limited risk groups [8,11,24,26,29], and two compared with a control group where routine culture-based screening was performed [14,18]. Among the eight studies which could not document decreasing trends in MRSA infections or transmission following the implementation of screening, three compared PCRbased screening with culture-based screening [7,13,23], four compared the intervention to control periods without any active surveillance of MRSA [17,22,39,41], and one compared the intervention with a baseline period where PCR-based screening of selected risk patients was performed [25].

PCR-based vs culture-based screening
Five investigations compared PCR-based to culturebased screening [7,13,14,18,23]. All five documented that the TAT was reduced when compared to culturebased approaches (Table 1). However, three studies found no difference in MRSA acquisition or infection rates [7,13,23]. In contrast, one before-and-after study found a reduction in the incidence of MRSA transmission after introduction of the PCR-based test which almost reached statistical significance, and one cohort study reported a reduction in MRSA acquisition rates [14,18].

Decolonisation
A total of 11 RCTs, 23 comparative studies and one prospective cohort study evaluated the effectiveness of mupirocin-based nasal decontamination regimens for the prevention of S. aureus infections (  [48,51,52,72,73,75]; for one of these, this was only observed when selective digestive decontamination was added to nasal decolonisation [52], and for one RCT, the effect was only analysed for Gram-positive infections (which were mostly MRSA) [75]. Stratified by types on infections prevented, the RCTs showed that decolonisation decreased deep S. aureus SSI [48], overall S. aureus infections [48,51,73], overall infection rates [52], Gram-positive pneumonia [75] and S. aureus exit-site infections [72].

Isolation
Focusing on the physical isolation of patients in separate single or cohort rooms, we identified one cohort study and seven comparative studies reporting on the effectiveness of this measure (Table 3) [16,[83][84][85][86][87][88][89]. Five studies were performed in ICU settings [16,[83][84][85]88], one in a vascular surgery ward, one in a diabetic food unit, and one hospital-wide [86,87,89]. In two of these studies, nurse cohorting was performed in addition to single-room isolation [83,86]. Overall, one cohort and three comparative studies reported on beneficial effects of single-room isolation (not performed pre-emptively) on MRSA colonisation or infection [85,86,88] and on acquisition rates [84]. Two comparative studies did not find a reduction of transmission [83] or MRSA prevalence [87].
Three studies assessed the role of pre-emptive isolation measures pending the results of screening [16,86,89]. In one before-and-after study, pre-emptive isolation precautions led to a reduction of the MRSA acquisition rate (0.21% vs 0.07%; p=0.04) [89]. In a retrospective comparative study placing all admitted patients in pre-emptive isolation, the number of nosocomial MRSA isolates was reduced (p=0.005). However, simultaneous introduction of a cohort isolation facility with dedicated staff makes the effects of this measure indistinguishable from the effects of preemptive isolation [86]. The third was a study that evaluated the effects of simultaneous implementation of pre-emptive isolation and a rapid screening test on the incidence of MRSA infections in two ICUs [16] resulting in a significant reduction of ICU-acquired infections in a medical but not in a surgical ICU.

Discussion
Improving the rational use of antibiotics and the implementation of hand hygiene are clearly cornerstones of MRSA prevention and control [90][91][92]. Moreover, benchmarking and public reporting systems have recently been demonstrated to successfully support infection control measures [93]. However, the effectiveness of screening, decolonisation and isolation for MRSA prevention when implemented routinely in settings with endemic MRSA, remains controversial. For example, it is debated to what extent microbiological, strain-specific factors have contributed to the decreasing MRSA trends [94,95]. Therefore, the present review aimed to focus on three important measures and to summarise the current evidence for their impact on MRSA prevention.

Screening
The strategy of screening is based on the finding that microbiological cultures performed for clinical reasons fail to detect previously unknown MRSA carriers at admission in 69 to 85% of patients [96,97]. Technically, screening can be performed by culture-based methods (screening swab streaked onto non-selective or chromogenic media) or PCR-based tests.     [98]. On the other hand, two RCTs found that MRSA acquisition or infection in the intervention groups did not differ significantly from the control groups [9,20]. However, in both studies, the median time for reporting a positive screening result was very long (3 days and 5.2±1.4 days), which led to delayed implementation of contact precautions. In addition, compliance with transmission-based precautions was not as required [20] and the prevalence of MRSA infection was low in one of the studies [9]. Comparing successful and unsuccessful interventions, we did not find clear differences between the studies regarding the specimens used for screening (nasal swab only vs other swabs in addition) or the patient population included (all patients admitted vs high-risk patients only).
There was a tendency that studies including 'incidence of MRSA acquisition' as an outcome parameter, reported a success less frequently (three of eight studies) compared with studies focusing on MRSA infection rates using the outcome parameters 'occurrence of bacteraemia' (eight of nine studies) or 'SSI' (three of five studies). The reason for this effect is not known, but it could highlight that screening does not necessarily affect the rate of cross-transmission on the ward, unless it is linked to additional preventive measures; decolonisation, for instance, was not performed in two of the the studies measuring incidence of acquisition [33,34], while in two others, single-room isolation was omitted or only performed if available [9,17].
In conclusion, we found evidence that screening can help decrease MRSA infection rates in hospitals. This is also supported by macro-epidemiological data and mathematical models showing that without screening, other infection control measures might fail to effectively reduce MRSA spread [99][100][101][102]. However, the included RCTs did not confirm the findings of noncontrolled studies. This makes it impossible to firmly recommend the implementation of screening in all settings. However, the evidence provided can support the introduction of a programme for active surveillance of MRSA in settings that have hyperendemic MRSA crossinfections in spite of a high level of compliance with standard precautions. Clearly, the implementation of screening needs to be linked to other targeted infection control measures (e.g. hand hygiene) to achieve optimal impact.

Culture-based screening vs PCR-based screening
Screening for MRSA colonisation of patients at admission using culture-based approaches requires 24 to Table 2g Studies  72 hours until the results are available on the wards [103,104]. During this time MRSA can spread among inpatients. Therefore, various PCR-based methods have been developed to reduce the TAT [105,106]. Reduction of TAT was indeed confirmed by all studies on PCR-based tests identified in this review. But these studies mostly did not find a significant reduction of MRSA infection or acquisition rates. These results are in accordance with data from a meta-analysis showing that, compared with cultures, the use of rapid tests was not associated with a significant decrease in MRSA acquisition rates (risk ratio 0.87; 95% CI: 0.61-1.24) [98]. On the other hand, we found two studies reporting on a significant reduction of MRSA acquisition and a trend towards declining transmission [14,18]. They demonstrate that implementation of PCR-based surveillance can be beneficial at least in facilities where culture results have a very long TAT (>3 days) [14,18].
We conclude that in settings where MRSA screening based on cultures, followed by the implementation of additional precautions, is already implemented, the current evidence does not suggest replacing or supplementing culture-based surveillance with rapid tests. However, besides accelerating the implementation of additional precautions, the high negative predictive value of MRSA rapid tests may also be useful when discontinuing contact precautions (including single-room isolation) in settings where they are implemented preemptively for suspected MRSA carriers [103]. However, the reliability of a negative nasal rapid test has not been evaluated in situations where pre-emptive isolation is performed for high-risk patients, who are often carrying MRSA at extranasal sites (e.g. wounds). Furthermore, using rapid tests in low prevalence settings may increase the number of false-positive tests (positive predictive values: 31-78%) [103,[107][108][109][110].

Decolonisation
The effectiveness of mupirocin nasal ointment to eradicate MRSA has been estimated to be 94% one week after treatment and 65% after a 14-day followup period [111,112]. Effectiveness of MRSA decolonisation therapy is obviously limited when extranasal sites are colonised [113]. Since nasal carriage of S. aureus is a major risk factor for subsequent nosocomial infection, there is a theoretical rationale that eradicating S. aureus from the nares can reduce the development of infection. It is, however, controversial to what extent studies assessing the effectiveness of decolonisation among patients carrying MSSA also hold lessons for MRSA [114]. In this review, we have identified only four studies in which mupirocin-treatment was not restricted to MRSA carriers and where effects on MRSA and MSSA infections were reported separately. All four documented a decrease in MRSA, but found insignificant results for MSSA [64,77,79,82]. However, this does not mean that mupirocin-based decolonisation is ineffective against MSSA in general, since two randomised trials have reported a reduction of MSSA infections [48,51]. The reasons for this discrepancy are Table 3b Studies unknown, and the question whether results obtained for MSSA can be transferred to MRSA is unresolved. Despite potential local differences in mupirocin susceptibility and the occurrence of clonal lineages [114], a plausible biological explanation why results on MSSA decolonisation treatment should not be applied for MRSA, is currently lacking. Therefore, we have explicitly included studies dealing with S. aureus decolonisation. However, future studies will have to assess in detail the differences between the preventive effectiveness of MSSA and MRSA decolonisation.
Overall, we conclude that, taking into account local rates of healthcare-associated infections and infection control conditions, mupirocin-based decolonisation therapy should be considered for selected S. aureus carriers who are at high risk of developing nosocomial S. aureus infections. The best evidence is available for patients undergoing cardiothoracic or orthopaedic surgery. Of note, the preventive use of mupirocin for decolonisation is constrained by the development of resistance, found in 1% of all subjects when mupirocin was used for short-term prophylaxis. Increasing lowlevel mupirocin resistance (8-256 µg/mL) has recently been reported in parallel to increased mupirocin consumption [112,115,116].

Isolation
There are multiple approaches to organise isolation measures: Patients can be transferred to special isolation wards, housed in nursing cohorts with designated staff, isolated in single or cohort rooms on general wards without designated personnel, or housed in the same room as patients not affected by MRSA while applying barrier precautions (e.g. gloves and gowns) when caring for the MRSA patient. In this review, we focussed on single room or cohort room isolation because this measure is sometimes debated as it can be associated with disadvantages for the isolated patient [117]. Moreover, in settings with a high prevalence of MRSA, isolation of patients may be hindered due to insufficient side room capacity and financial constraints, if isolation results in bed-blocking.
Overall, we found four studies showing that single room isolation led to a reduction in nosocomial MRSA acquisition and in the incidence of MRSA infection [84][85][86]88]. In contrast, in a prospective interrupted-timeseries study it was found that, MRSA acquisition was not different in phases during which MRSA-colonised or infected patients were moved to single or cohort isolation, compared with phases during which they were not moved [83]. However, limitations of this study are delayed notification of screening results, a high number of missed screenings (80-87% of patients at admission and 71-75% at discharge) and low compliance with hand hygiene (21% compliance) [83]. Moreover, a retrospective comparative study showed that discontinuing single-room isolation and applying transmission-based precautions (e.g. masks, gowns, gloves) for MRSA patients did not lead to an increase in the prevalence of MRSA. However, that study did not measure the occurrence of transmission on the wards and the incidence of MRSA infections [87].
We conclude that the limited evidence from non-controlled studies which is available to support the use of single-room isolation for MRSA (outside of outbreaks) should inspire further research in this field to facilitate the development of evidence-based guidance in future, also for the prevention and control of other multidrugresistant organisms. However, the majority of studies identified and observations made during outbreaks support the use of single-rooms [3]. Therefore, where facilities (isolation wards, single rooms, cohort rooms) for the isolation of MRSA patients are available, their use should be recommended.
In all investigations identified, it is difficult to estimate to what extent the observed preventive effects were attributable to pre-emptive isolation or to other measures implemented in parallel [16,86,89]. Consequently, there is a need to assess the evidence for the use of pre-emptive isolation measures in hospitals. This is of major importance, because authors evaluating PCR-based screening tests often suggested that rapid tests could accelerate the start of isolation precautions [16,103,118]. However, these advantages cannot be assessed adequately as long as the additional value of pre-emptive isolation is unclear.

Conclusion
We have documented that the evidence for the effectiveness of three major MRSA prevention and control measures does not allow for clear guidance offering 'one-size-fits-all' solutions, because the effectiveness of these interventions seems highly depending on the prevalence of MRSA, compliance with general infection control measures (e.g. hand hygiene), the incidence and type of infections and the transmission rates within the respective setting of implementation. This is documented by the ambiguous study results presented here. In addition, models on the effectiveness of MRSA prevention strategies in different settings have shown that even measures which are performed highly effectively in outbreaks or low-prevalence areas, failed to control MRSA when applied for long-term control or in high-prevalence settings [119]. These difficulties have led to the development of models describing the effects and costs associated with universal vs selective MRSA screening in different settings, which may facilitate the implementation of local standards [104,120]. Moreover, some authors have recently described the effectiveness of several preventive bundles comprising the measures reviewed here in combination with other interventions. For example, it was shown that universal nasal screening, contact precautions for patients colonised or infected with MRSA, hand hygiene, and changes in the institutional culture of responsibility reduced MRSA infections by 62% [99].
Others have identified that structural factors such as engaging front-line staff, building multidisciplinary teams, providing monitoring and feedback, and acquiring management support were key measures for the success of MRSA prevention [121]. The evaluation of such bundles with respect to their effects, feasibility and applicability in different healthcare systems (e.g. different countries), clinical departments and patient collectives could in the future guide preventive efforts. Compared to assessing the effects of single preventive measures separately (as done in this review), the main advantage of assessing the effects of bundles is that they are planned specifically for targeted healthcare sectors, and the assessment can take into account the financial and other structural conditions in the respective settings.
In this review, we did not restrict the eligibility criteria to controlled studies such as RCTs, although quasiexperimental study designs are prone to be associated with various biases (e.g. selection bias or size of study population). This was done because only very few controlled investigations have been published. In addition, among the 14 RCTs included, most of which were performed for assessing the effectiveness of decolonisation therapy, a majority did either include patients affected by MSSA or did not stratify their outcomes for MSSA and MRSA infections. This makes the results, even of these formally 'high-quality' studies, disputable. Against this background, we decided not to perform a formal grading of the quality of the included studies, but rather to present the study results holistically and leave their use in various settings and countries open for interpretation.
The controversy about different implementation pathways for screening, isolation and decolonisation should not obscure the fact that the beneficial effects of MRSA control measures in general [120] support the recommendations made in many European national MRSA policies from low prevalence countries (e.g. the Nordic countries and the Netherlands) and high prevalence countries (e.g. France, Germany, and the United Kingdom), where a combination of these measures are the standard of care and a reduction in MRSA infections has recently been achieved by coordinated efforts even in high prevalence settings [5,122].