How much patient-to-patient spread of S. aureus occurs? Apparently, not much

I recently posted an article on surprising finds of a study suggesting that horizontal transmission of C. difficile from known symptomatic cases may be less common that we thought. A group of researchers from Oxford, Brighton and London in the UK applied similar methodology to Staphylococcus aureus transmission with similar findings: only a fifth of S. aureus acquisitions could be attributed to patient-to-patient transmission.

All patients admitted to a 16 bed ICU in Brighton were screened on admission and weekly to detect S. aureus colonization and acquisition. Each isolate was typed by spa and whole genome sequencing (WGS). The number of acquisitions that could be linked using conventional methods (spa typing combined with an analysis of overlapping stays) vs. WGS was evaluated.

Overall, 185 (16.7%) of 1109 admissions carried S. aureus; 59 carried MRSA (5.3%). 680 patients were on the unit for long enough to have a weekly screen and hence were eligible for assessing acquisition. Of these, 44 S. aureus (22 MRSA) acquisitions were detected in 41 patients. 35 of these acquisitions were in patients who were screen-negative on admission and 9 were acquisitions of different strains in patients who were already colonized on admission.

Only 14% (5/36) of the acquisitions available for typing were of the same spa type as another patient on the unit at the same time. All of these were MRSA. WGS discounted three of these apparent occurrences of patient-to-patient transmission, confirmed two and identified a further five (3 MRSA, 2 MSSA). So, in total, 7 / 36 (19%) of acquired isolates (5 MRSA and 2 MSSA) were linked to isolates from other patients (Figure).

Price MRSA WGSFigure: Source of S. aureus acquisitions identified through WGS.

The paper raises some interesting questions:

  • Principally, if almost 80% of patients did not acquire their S. aureus from other patients on the unit, where on earth did they acquire them from? In the C. difficile study, unrecognized importation of diverse strains from the community that would not be detected on admission since admission screening was not performed represents a plausible explanation for the surprisingly low incidence of horizontal transmission from known cases. This is not the case in this study, where all patients were screened on admission for S. aureus. So where were the diverse isolates acquired from? Staff carriers? Contaminated surfaces? S. aureus has the capacity to survive on surfaces for very long periods (more than 1 year), so an ancient environmental reservoir is possible. Furthermore, there was no ‘lead in’ period to the study, so it could be that S. aureus on the unit in the months before the study left an environmental reservoir that led to acquisition in some cases, which would not have been captured by this study. The fact that 4/7 patient-to-patient transmissions acquired the same strain of S. aureus without sharing time on the ICU together supports this, although could also be explained by staff carriers. It was a shame that 16% of the 329 S. aureus isolates (including 16% (7/44) of the acquired isolates) were not available for sequencing, which represents a substantial reservoir from which patient-to-patient transmission likely occurred in some cases.
  • It was interesting that some 20% (9/44) acquisitions that were detected occurred in patients who were already colonized on admission; these would have been missed altogether if molecular typing was not performed. I wonder how much ‘silent’ acquisition of this type occurs?
  • Assuming temporal relationship between strains assumes a constant mutation rate. The ‘speed of the mutation clock’ was assessed in this study through repeated sampling of the same patient. This exercise demonstrated minimal diversity, as was the case for C. difficile.
  • WGS is rapidly becoming the gold standard for transmission mapping. In this study, the conventional approach of evaluating spa typing with overlapping stays lacked both sensitivity and specificity for identifying transmitted isolates.

In summary, the major finding is that only 20% of patients who acquired S. aureus appeared to acquire it through horizontal spread from other patients. The next frontier of transmission mapping must be a more comprehensive evaluation of other potential sources: contaminated surfaces, contaminated air, nurses, doctors, cleaners, tea-tray deliverers and the list goes on…

Article citation: Price JR, Golubchik T, Cole K, Wilson DJ, Crook DW, Thwaites GE, Bowden R, Walker AS, Peto TE, Paul J, Llewelyn MJ. Whole-genome sequencing shows that patient-to-patient transmission rarely accounts for acquisition of Staphylococcus aureus in an intensive care unit. Clin Infect Dis 2014. Jan 9. [Epub ahead of print].

High levels of antibiotic resistance and low levels of ‘antibiotic literacy’ in the general public are on a dangerous collision course

MRSAPhoto credit: CDC, Janice Haney Car.

Guest Blogger Dr. Rodney E. Rohde (bio below) writes: The U.S. Department of Health and Human Services (HHS) has identified the reduction of healthcare-associated infections (HAIs) as an Agency Priority Goal for the Department. HHS is committed to reducing the national rate of HAIs by demonstrating significant, quantitative, and measurable reductions in hospital-acquired central line-associated bloodstream infections and catheter-associated urinary tract infections (1). In this national context, recent data show that Americans visit the doctor approximately 12 million times each year to get checked for suspected staphylococci or MRSA skin infections. Estimates are that 1 of every 20 hospital inpatients will contract a HAI (2).  This is roughly 270 deaths per day by one estimate – think of an airplane disaster with no survivors every day as a comparison. It appears that more people in the US now die from MRSA, especially healthcare acquired, than from HIV/AIDS.

MRSA incidence and the low levels of literacy about antibiotic resistant infections in the general public are on a natural collision course. The health implications for society are startling and this brewing public health storm should be a wake-up call for all involved in the prevention and treatment of MRSA and other antibiotic resistant infections. Traditional medical approaches to infection control, antibiotic prescribing and usage, and how the healthcare practitioner relates to the public in general, and to individuals in particular, are no longer adequate in light of this growing healthcare emergency. It has been my experience that individuals who are diagnosed with MRSA infections (and other HAIs) need an informed healthcare professional to explain a number of complicated issues to them to help the individual understand the dangers of these nasty antibiotic resistant infections. This, in turn, will lead to greater acceptance and adoption of infection control and prevention.

For example, I conducted a study for my dissertation that details the experiences of ten participants who have been diagnosed with MRSA (3). Briefly, this study provides a new model of how a person with MRSA in the general public learns and adapts to the infection. Their experiences emerged to create critical implications for practice and research. Almost all individuals had important advice for professionals about the nature of a consistent message for an MRSA plan. The embedded features of this plan were that people make the difference in these life-changing diagnoses. Likewise, all participants echoed the need for consistency of MRSA information and a step-by-step plan to manage the condition. Particular attention should be paid to the following areas: (a) the patient-healthcare provider interaction, specifically to create an open and non-threatening environment for learning to occur, (b) the delivery of critical information about the importance of having an MRSA diagnosis based on laboratory culture and antibiotic susceptibility testing, (c) specific education on what a MRSA infection looks like, including images/pictures and MRSA stories for patients, (d) the use of podcasts, digital video, and other electronic media (e.g. Facebook) to provide patient education beyond the initial MRSA diagnosis, (e) specific education about infection care, control, and prevention to themselves and others, and (f) guidance for individuals about sources of information and the credibility of sources. In regard to the use of podcasts and other electronic media, the message should be formed with a combination of stories from MRSA survivors and healthcare professionals to build a strong, synergistic media tool.

The implications for practice and research based on the literature and results of this study indicate a need to address issues of how the general public discovers, learns, and adapts to antibiotic resistant infections, especially MRSA. Likewise, this study emphasizes the critical importance of informing healthcare professionals and health educators (e.g. universities, schools, and other related institutions) about the need for better programs of patient education and continuing education surrounding the pre and post diagnosis of MRSA infections.  The participants in this study all emphasized the critical nature of talking to others that had already experienced MRSA with respect to getting an insider’s perspective on lessons learned. Finally, the participants in this study appeared to utilize self-directed learning and, to a lesser extent, transformational learning to challenge the healthcare system in regards to what content they needed about MRSA and how they might best learn to understand the disease. It may be possible to build on this desire to impact the healthcare system by inviting individuals who have experiences with MRSA to join a healthcare associated advisory committee.

References

  1. US Department of Health and Human Services, Health Care-Associated Infections.
  2. Department of Health and Human Services. HHS Action Plan to Prevent HealthCare-Associated Infections: Roadmap to Elimination.US Department of Health and Human Services, Washington, DC, USA (2012).
  3. Rohde, R.E. & Ross-Gordon, Jovita. MRSA model of learning and adaptation: a qualitative study among the general public. BMC Health Services Research, 2012, 12:88.

Guest Blogger Bio

rohde

Dr. Rodney E. Rohde is a Professor, Research Dean and Program Chair of the Clinical Laboratory Science program in the College of Health Professions, at Texas State University.

Dr. Rohde’s background is in public health and clinical microbiology. He has a bachelor’s degree in microbiology, a master’s degree in biology/virology and a PhD in education from Texas State. His dissertation was aligned with his clinical background: MRSA knowledge, learning and adaptation.

His research focuses on adult education and public health microbiology with respect to rabies virology, oral rabies wildlife vaccination, antibiotic resistant bacteria, and molecular diagnostics/biotechnology. He has published over 30 research articles and abstracts and presented at over 100 international, national and state conferences. He was awarded the 2012 Distinguished Author Award and the 2007 ASCLS Scientific Research Award for his work with MRSA. Recently, his work was the focus of an educational campaign regarding the important research focus of MRSA, which featured Dr. Rohde in a video by Texas State University that has been used by numerous media outlets. Learn more about his work here.

This study has been BUGGing me for a while

bug glove

A fabulous study recently published in JAMA evaluates the ‘Benefits of Universal Glove and Gown’ (BUGG) in US ICUs. This is a model study design: one of the first cluster randomized controlled trials of a non-therapeutic infection control intervention. Twenty ICUs were paired and randomized to either universal glove and gowning, or to continue the current practice of placing patients known to be infected or colonized with MRSA and VRE on contact precautions. The hypothesis is that undetected colonization with MRSA and VRE is common, and the only real way to address this is to assume everybody is colonized!

Summary of findings:

  • Universal glove and gowning was not associated with a reduction in a composite measure of MRSA / VRE acquisition (the primary outcome).
  • VRE acquisition was not reduced by universal glove and gown use, whereas MRSA was.
  • CLABSI, CAUTI and VAP; ICU mortality; and adverse events did differ significantly between the two groups.
  • Hand hygiene compliance on room entry was not significantly different between the two arms, whereas hand hygiene compliance on room exit was significantly higher in the intervention arm.
  • Healthcare workers visited patients 20% less frequently in the intervention arm (4.2 vs. 5.2 visits per hour).

BUGGFigure: The change in acquisition rate, comparing the baseline period with the study period for the intervention and control units.

Here’s what’s BUGGing me about this study:

  • The acquisition rate in both intervention and control study arms reduced (Figure). The acquisition rate reduction in the control arms may be due to improved compliance with admission screening, resulting in more accurate ascertainment of who required contact precautions.
  • The significant reduction was achieved for MRSA but not for VRE. The authors suggest that VRE colonization may have been suppressed on admission and not detected, and flourished during antimicrobial therapy giving the impressive of acquisition. I wonder whether differences in the routes of transmission may also have contributed; for example, VRE seems to be substantially “more environmental” than MRSA. Another potential confounder is that, by chance, the prevalence of MRSA or VRE on admission to the intervention ICUs was more than double that in the control ICUs (22% vs. 9%). In actual fact, the raw rate of MRSA acquisition in the intervention ICUs was marginally higher than in the control ICUs during the intervention period (6.00 vs. 5.94 per 1000 patient days), even though the change in rate was significantly greater on the intervention ICU. Although adjustment was made for this difference in the analysis, it may have skewed the findings somewhat.
  • The authors achieved remarkably high compliance with admission screening (around 95%), discharge screening (around 85%) and glove and gowning (around 85%). Each site had the luxury of a study coordinator and a physician champion to lead implementation, plus weekly feedback on screening compliance and visits from study investigators. Most ICUs would not be afforded these luxuries so I suspect that real-world compliance outside of the somewhat artificial study environment would be considerably lower. Indeed, an ID Week poster suggests that compliance with gowning in one US ICU was a ‘dismal’ 20%!
  • Adverse events were not significantly higher in the universal glove and gowning arm, which may seem surprising prima facie. However, the reason why adverse events are more common for patients on contact precautions is that they are marginalized by being on contact precautions. If all patients are effectively on contact precautions, the time of healthcare workers would be spread evenly.
  • Universal gloving is likely to result in universally bad hand hygiene compliance within the room during patient care; when healthcare workers feel protected, they are less likely to comply with hand hygiene and gloves are a good way to make healthcare workers feel protected. The increase in hand hygiene compliance on room exit is probably also a symptom of inherent human factors, since healthcare workers feel more ‘dirty’ when exiting the room of a patient with a higher perceived risk of MDRO ‘contamination’ (the so-called “urgh” factor).
  • Healthcare workers had less time for patient care in the intervention arm because they were busy donning and doffing gloves and gowns. Interestingly, the authors suggest that fewer visits may be a good thing for patients, and may have contributed to their reduced chances of acquiring MRSA. This seems unlikely though, given the fact that VRE acquisition was not reduced. On balance, less contact with healthcare workers is likely to be bad for patients.
  • The increased cost of universal glove and gowning was not evaluated and, whilst incrementally small, would be a substantial sum.

In summary, this study sets the standard in terms of rigorous assessment of an infection prevention and control intervention. Universal application of gloves and gowns is unlikely to do as much harm as universal administration of mupirocin, but it will not make a profound reduction in the transmission of MDROs. Therefore, I shouldn’t think many ICUs will be rushing to implement universal gloves and gowns on the strength of these findings.

Article citation: Harris AD, Pineles L, Belton B et al. Universal glove and gown use and acquisition of antibiotic-resistant bacteria in the ICU: a randomized trial. JAMA 2013;310:1571-1580.

Is “community-acquired” CDI real?

A recent high profile US study delved into apparent community-associated CDI cases to evaluate healthcare exposures. The study was large, evaluating almost 1000 cases of community-associated CDI from 8 US states. Only 177 (18%) of the 984 cases had no recent healthcare exposure (Figure 1). Furthermore, healthcare exposure was only evaluated for the 12 weeks prior to the positive specimen, so I would wager that a portion of this 18% acquired their infecting C. difficile in a healthcare facility.  CA-CDIFigure 1. Data demonstrating that most MRSA and CDI presenting on admission to hospital are likely to have been acquired in a healthcare facility.  

So, it seems that the majority of these cases are more likely to be community-onset, healthcare-acquired CDI, rather than community-acquired CDI. I feel like we’ve been here before. In the 1990s before the emergence of distinct strains of CA-MRSA, MRSA presenting at hospital admission was commonly termed ‘community-associated’ or, worse, ‘community-acquired’ when really it was MRSA that had been acquired in hospital during a previous stay (Figure 1). The situation has now changed since distinct MRSA clones have emerged that have the capacity to cause infection outside the healthcare environment.

Turning our attention to the UK, the mandatory report scheme classifies cases of CDI as ‘Trust-apportioned’ if the specimens is collected from patients who have been in hospital for four or more days (Figure 2). It is tempting to speculate that the cases of CDI that are non Trust-apportioned are CA-CDI. However, the definition for ‘Trust-apportioned’ does not account for previous healthcare contact, and the rate of Trust-apportioned and non-Trust-apportioned cases tracks so closely that, once again, these are likely to be healthcare-acquired CDI presenting on admission.

Slide1Figure 2. Number of cases of CDI in England through the mandatory reporting scheme, 2004-2013.

The epidemiology of C. difficile is fundamentally different to MRSA, in that healthy neonates typically have a high rate of C. difficile colonization. Thus, there is a ready reservoir for a low rate of genuinely community-acquired CDI. However, it seems to me that most “CA-CDI” reported thus are likely to be acquired in a healthcare facility and I have not seen any data to convince me that community-acquired CDI is increasing.

Article citation: Chitnis et al. Epidemiology of Community-Associated Clostridium difficile Infection, 2009 Through 2011. JAMA Intern Med 2013;173:1359-67.

The pitfalls of PCR for detecting pathogens on surfaces

PCR has proven an invaluable tool for the rapid diagnosis of a range of pathogens, including MRSA and C. difficile. Several studies have evaluated the potential use of PCR for the detection of pathogens on surfaces and have identified some issues that, frankly, seem pretty terminal for this application using currently available commercial PCR kits.

A study from Cleveland evaluated the use of a commercial RT-PCR test for detecting C. difficile on hospital surfaces. Three composite sites were sampled in 22 patient rooms, 41% of which housed a patient with CDI with the remaining 59% sampled after terminal cleaning and disinfection. Two swabs and a gauze were collected from each site; one swab was cultured directly onto selective agar and the other was tested using PCR. The gauze was cultured using broth enrichment. C. difficile that grew on the selective agar were tested for toxin production and only toxigenic C. difficile were included.

Overall, 23 (35%) of the 66 sites grew toxigenic C. difficile and only 4 of these were detected using the standard RT-PCR assay (sensitivity 17%, specificity 100%). The sensitivity of RT-PCR in rooms that had been cleaned and disinfected was even worse (10%). Increasing the CT threshold of the assay (making it less stringent) improved the overall sensitivity to 52% and did not affect the specificity.

The study has several important limitations. The RT-PCR assay detected only the Toxin B gene, whereas the toxigenic culture methodology would detect both Toxin A and B producers. More importantly, there was a crucial difference in sampling methodology: the gauzes used for broth enrichment culture had a 50% higher positivity rate than the swabs (in line with other findings), but only swabs were tested by both PCR and culture. Thus, if the gauzes are a more effective sampling device, this would make the RT-PCR methodology seems worse than it is. I would have liked to have seen the sensitivity of the RT-PCR assay for detecting C. difficile cultured from the swabs only, but I could not derive this from the data in the paper.

An older study from New Haven, Connecticut provides a contrasting view of the use of PCR to detect pathogens from surfaces. Here, 10 standardized sites were sampled in the rooms of 10 patients infected or colonized with MRSA, and 5 rooms of patients not known to be infected or colonized with MRSA. Swabs were directly plated onto selective agar for MRSA, then DNA was extracted from the swabs before a broth enrichment procedure using the same swabs. In this study, 40 (27%) of the 150 surfaces were positive by culture, but 90 (60%) were positive by PCR (sensitivity 93%, specificity 51%).

Deshpande 2013

Figure 1. Contrasting sensitivity and specificity when using PCR to detect C. difficile and MRSA on hospital surfaces.

It seems then that the sensitivity of PCR is too low for the environmental detection of C. difficile but the specificity is too low MRSA (figure 1). How could this be? Assuming that this is not due to experimental differences between the studies, it could be that the standard extraction procedure used for the C. difficile assay was not robust enough to liberate DNA from the mature environmental spores, resulting in low sensitivity. Conversely, the PCR assay was detecting DNA from dead MRSA on surfaces, resulting in low specificity.

So, in summary, the MRSA assay was too sensitive and the C. difficile assay was not sensitive enough! While the use of these “off the shelf” commercial assays doesn’t seem to be useful for detecting pathogens on surfaces, there may be hope for a PCR assay tailored specifically for an environmental application.

Article citations:

Deshpande A, Kundrapu S, Sunkesula VC, Cadnum JL, Fertelli D, Donskey CJ. Evaluation of a commercial real-time polymerase chain reaction assay for detection of environmental contamination with Clostridium difficile. J Hosp Infect 2013;85:76-78.

Otter JA, Havill NL, Boyce JM. Evaluation of real-time polymerase chain reaction for the detection of methicillin-resistant Staphylococcus aureus on environmental surfaces. Infect Control Hosp Epidemiol 2007;28:1003-1005.

The effect of closing and cleaning wards on infection rates

Not so long ago, the UK Government ordered a national ‘deep clean’. This prompted a fair amount of debate among experts and the public. If the NHS needed a spring clean, then does that mean that it was dirty in the first place? Perhaps. There does not seem to have been a formal evaluation of impact, but there is some rationale for closing and cleaning wards. For example, this paper from the early 1970s evaluated the impact of closing and cleaning five wards in London.

The five wards (four surgical and one medical) had an outbreak of MRSA (termed ‘cloxacillin-resistant S. aureus’). Rates of infection (termed ‘sepsis’) were monitored on the study wards before and after closing and cleaning. Wards were closed to admissions and emptied of patients. All fabrics were sent for laundering and all left over supplies were discarded. Cleaning comprised washing floors, walls and all other surfaces with hot water containing detergent; bed frames and furniture were also washed. The length of time that all this cleaning too is not specified, but I suspect it took place over several days. Crucially, staff and patients were screened for carriage of epidemic strains of S. aureus; colonised patients were not re-admitted after ward cleaning where possible.

The charts below show the impact on all infections (Figure 1), all S. aureus infection (Figure 2) and MRSA infection (Figure 3). Infection rates were compared 3 months before vs. 3 months after cleaning on Wards 1-3 and 6 months before vs. 6 months after on Wards 4 and 5. As you can see, the impact was pretty dramatic.

Noone Fig 1

Figure 1. Total infection rate (proportion of admissions infected) on the five wards before vs. after ward closing and cleaning.

Noone Fig 2

Figure 2. S. aureus infection rate (proportion of admissions infected) on the five wards before vs. after ward closing and cleaning.

Noone Fig 3

Figure 3. MRSA infection rate (proportion of admissions infected) on the five wards before vs. after ward closing and cleaning.

The poor reduction in total infection rate on Ward 1, a gynecological ward, (Figure 1) is largely due to high Gram-negative infection rates before and after cleaning, most likely explained by endogenous urinary tract infections. Reductions in total infection rate and S. aureus infection rate appeared to be less on Wards 4 and 5, which could be influenced by the fact that rates were compared for 6 months pre and post ward closing and cleaning rather than 3 months on Wards 1-3. The impact of a one off environmental intervention is likely to diminish over time. It’s also interesting to note that the MRSA infections identified on Ward 5, a general surgical ward, after cleaning were due to a different strain of MRSA (determined by phage typing and antibiogram) than before cleaning. This new strain matched the outbreak strain from Ward 2. Two of the patients on Ward 5 who became infected with this strain were operated on in the same theatre as the infected patients from Ward 2 within two weeks of one another. Four other patients (on different wards) also appeared to acquire the strain in the same operating theatre.

The study has several important limitations. It is not possible to be certain whether active screening and isolation or ward closing and cleaning were responsible for the reduction in infection rates; it was probably combined impact. The study design lacked the rigor of more modern investigations: infection rates were not expressed in terms of patient-days and infection rates were compared for different time periods making direct comparison of the impact across the five wards difficult. Also, no environmental sampling was conducted to demonstrate the efficacy of the cleaning procedure (both initially and in terms of recontamination).

Notwithstanding these limitations, the study provides evidence that ward closing and cleaning combined with active screening and selective readmission resulted in a dramatic reduction in the rate of nosocomial infection on five study wards. The impact appeared to be most pronounced in the first three months, which is consistent with a reduction in environmental contamination. Outbreaks of MRSA were eradicated by closing and cleaning on all five study wards. However, there was evidence of new nosocomial transmission following the re-admission of infected patients. Finally there was some interesting circumstantial evidence of transmission within operating theatres.

Article citation: Noone P, Griffiths RJ. The effect of sepsis rates of closing and cleaning hospital wards. J Clin Pathol 1971;24:721-725.

Universal MRSA decolonization will tend towards universal resistance

swabA seminal study recently published in the New England Journal of Medicine evaluated the impact of universal decolonization vs. targeted screening and / or decolonization to prevent ICU infection. The study has the potential to fundamentally change practice in ICUs around the world. However, I have two major problems with the study, namely mupirocin and chlorhexidine.

Study design and results. The authors describe their design as a ‘pragmatic’ randomized controlled trial, which should serve as a model for other studies. There were several variables that the authors could not change, for example, the fact that some US states mandate universal screening whereas others do not. They were able to cleverly incorporate this into their stratified randomization process so that the integrity of the study was not compromised.

The study was large, with 74,256 patients admitted to 74 ICUs in 43 hospitals randomized to the following groups during the intervention phase:

Group 1: Targeted screening and isolation.

Group 2: Targeted screening, isolation and decolonization (using intranasal mupirocin and chlorhexidine bathing).

Group 3: Universal decolonization (using intranasal mupirocin and chlorhexidine bathing).

The study compared the hazard ratio for MRSA clinical cultures, all-pathogen bloodstream infections and MRSA bloodstream infections in the 12 month pre-intervention vs. the 18 month intervention period for each of the groups. The main finding was that universal decolonization (Group 3) was associated with significantly lower rates of MRSA clinical cultures and all-pathogen bloodstream infection, and reduced rates of MRSA bloodstream infection (Chart). Target screening and isolation (Group 1) was the least effective strategy and targeted screening, isolation and decolonization (Group 2) had intermediate effectiveness (Chart).

Huang 2013

[Chart: hazard ratios and 95% confidence intervals comparing the pre-intervention with the intervention period in each of the three groups. Group 1: Targeted screening and isolation; Group 2: Targeted screening, isolation and decolonization; Group 3: Universal decolonization.]

Major problems: My major issue with the study is the universal (aka indiscriminate) use of mupirocin and, to a lesser extent, chlorhexidine. MRSA are able to acquire low-level resistance to mupirocin through mutations in ileS or high-level resistance through the acquisition of mupA. I can’t help thinking that the authors underestimate the risk of driving mupirocin resistance with their comment in the discussion: ‘Mupirocin resistance has been reported in some studies of MRSA decolonization, but not all such studies.’  It is frankly naive to think that universal use of mupirocin for each ICU admission will not result in the widespread development of mupriocin resistance. Thus, the comment in the accompanying editorial urging caution in implementing the universal use of mupirocin is warranted.

The situation is not as clear-cut with chlorhexidine. Clinically significant reduced susceptibility to chlorhexidine in S. aureus has been hard to define for a number of reasons:

  1. There is no generally agreed methodology for detecting reduced chlorhexidine susceptibility.
  2. Population MICs vary considerably making a meaningful breakpoint difficult to determine.
  3. Higher MICs displayed by some isolates (4–16 mg/L) are still well below the effective chlorhexidine concentration applied to skin (10000–40000 mg/L).
  4. Although genes encoding membrane efflux pumps have been identified in S. aureus, including qacA, these have an unclear relationship with reduced susceptibility.
  5. Other mechanisms of reduced susceptibility, such as decreased uptake of chlorhexidine, may confound the apparent effect of efflux pumps.
  6. Although many studies have reported on qacA carriage or reduced chlorhexidine susceptibility in S. aureus isolates, few have evaluated clinically relevant outcomes.

Notwithstanding these limitations, it is clear that the implementation of chlorhexidine based decolonization drives an increase in the carriage of genes associated with reduced susceptibility to chlorhexidine, and this may be clinically relevant based on studies from London and Geneva. For example, a recent article from my research lab in London found that the implementation of universal chlorhexidine-based decolonization was associated with the selection of qacA linked with a higher chlorhexidine MIC in one dominant endemic MRSA clone (CC22), but not another (CC30). The slower reduction in the CC22 MRSA bloodstream infection rate suggests that carriage of qacA confers a selective advantage.

So, will the introduction of universal decolonization result in resistance to mupirocin and chlorhexidine? The answer for mupirocin is a resounding yes; the answer for chlorhexidine is currently not clear.

Other limitations: It’s always easy to snipe at a high-profile study, but there do seem to be some important limitations. First, nasal screens alone were performed in patients admitted to Groups 1 and 2. The sensitivity of nasal screens alone can be as low as 50%, so the inclusion of other screening sites are preferable to improve sensitivity. The implications of missing many carriers on admission in Groups 1 and 2 are obvious.

Second, although the trend was in the same direction, there was no significant reduction on MRSA bloodstream infections for patients in Group 3. In fact, the crude rate of MRSA bloodstream infection was rather high in all three groups (0.5 to 0.7 per 1000 patient days in all groups during the baseline and intervention periods). For reference, the rate of MRSA bloodstream infection at Guy’s and St. Thomas’ NHS Foundation Trust in London was lower than this at the peak of the MRSA problems in 2003, approximately 10 times lower by 2008 and has fallen further since then. So, why the relatively high rate of MRSA bloodstream infection across the board in this study?

Third, while randomization is the “gold standard” in terms of study design, by chance, three of four hospitals that performed bone marrow and solid organ transplantation ended up in Group 3, which skewed this population somewhat. The study was large enough to “smooth out” this skew, but small studies that attempt randomization should take note. In a recent study with Johns Hopkins, we only had 6 units to play with so decided not to randomize but instead roughly “matched” the units in the two arms, which may be a better approach in smaller studies.

Fourth, it does not seem that any attempt was made to ensure that patients in Groups 1 and 2 were bathed daily. While bathing with chlorhexidine impregnated cloths has been shown to be superior to bathing with soap and water, it is not possible to determine whether daily bathing or chlorhexidine were most important.

Finally, compliance will never be 100%. Only 81% of patients in Group 3 actually received chlorhexidine and 86% received mupriocin, and only approximately 90% of patients in Group 2 that were due decolonization received chlorhexidine and mupriocin. However, the authors did achieve a very high compliance (in the high 90s) with admission screening in Groups 1 and 2.

Summary: The authors should be complimented for performing such a detailed and useful study. Unfortunately, the protocol was not configured to assess the relative contribution of mupirocin and chlorhexidine to the reductions in infection. Given the likely emergence of mupirocin resistance, future studies should drop the universal use of mupirocin all together and see how far universal chlorhexidine alone can go; recent data suggest it will go a long way. Indeed, the UK has managed to make a dramatic, national reduction in the rate of MRSA bloodstream infection without implementing universal mupirocin.

I have grave concerns about the universal use of mupirocin for all ICU admissions and, perish the thought, for all hospital admissions. To decide on implementing universal mupriocin use with a plan to keep an eye on resistance rates seems remiss (to put it politely) since the likelihood of widespread resistance is so high. I am less concerned about the universal use of chlorhexidine, although reduced susceptibility needs to be monitored carefully.

The accompanying editorial was entitled ‘Screening inpatients for MRSA – the case is closed’. For me, the only closed case emerging from this study is that universal, indiscriminate use of mupriocin is a bad idea. Antibiotics have caused the problem; they are not the solution.

Article citation: Huang et al. Targeted versus universal decolonization to prevent ICU infection. New Engl J Med 2013 May 29. [Epub ahead of print].