Tuesday, October 15, 2013

A good time to ditch contact precautions?

I get that no one likes to don gowns and gloves before seeing patients, particularly ID consultants who see a large number of patients colonized/infected with MDROs. I understand that touching is important for healing and that somehow wearing gloves impedes the healing touch. I even see why some still think that contact precautions place patients at greater risk for medical errors, even if the data supporting that contention is lacking. What I don't get is why infection prevention folks are pillorying one of the most effective methods we have in MDRO transmission prevention right when we need it the most.

My thoughts:
1) MDRO rates, particularly for Gram-negative pathogens like CRE and Acinetobacter are increasing
2) We have no antibiotics in our quiver, so prevention is our only hope for a decade+
3) Most prior room occupant studies are flawed and when proper methods are used, prior room occupants are not a risk factor - thus the environment isn't the only answer.
4) Hand hygiene compliance is only 60% - I know some are claiming 90%+ on their reports, but it's simply not true.
5) Gloves are massively effective in reducing the burden of organisms on hands and CP have now been shown to be effective in an RCT - (if you can look past the JAMA trial nihilists)

Which brings me to the table below modified from a study we completed a few years ago. You'll notice the red circle highlighting the per room entry contamination rate of healthcare workers hands with A. baumannii when hand hygiene compliance is 60%. The green circle highlights the contamination rate when gloves are worn with the 90% compliance achieved in the BUGG study (even ignoring that universal CP is associated with higher HH compliance).  You can see that when a healthcare worker enters an A. baumannii+ patient's room, 15% of the time they will leave the room with A. baumannii on their hands. If the hospital practices universal CP, it will be 3% of the time.

So, I get why people hate contact precautions, particularly ID-trained hospital epidemiologists. I just don't get the delight with which they limit their use given that CP are supported by the strongest evidence that our dismal science can muster. I suspect, again, that I won't have to live that long to witness their regret.


  1. Excellent post, Eli. I will respond with a counterpoint, in the spirit of the title of this blog!

    Of those who are changing the way they use contact precautions (CP), I don’t think I detect much delight. Rather, a thoughtful reassessment of how we use CP, not of whether CP should be used. The centers I referred to in my post yesterday are reserving CP for the exact organisms you mention (CRE, Acinetobacter) and scaling it back for those with colonization-only MRSA/VRE or ESBL-E. coli, and doing this based upon absence of data for benefit in the endemic non-ICU setting (and to emerging evidence that MDROs differ substantially in transmissibility, some of which evidence your group has provided). As to some of the other points you’ve made, I do think there is something in the prior room occupant literature, and at least one study that does appropriately control for colonization pressure. And I don’t think the BUGG assessment of adverse outcomes is at all dispositive about adverse effects of CP, at least not outside of ICU environments and for events that aren’t even designed to be captured by the IHI trigger tool (which include patient satisfaction concerns that, like it or not, are priority issues for hospitals).

    Regarding strong evidence for benefit of CP in non-outbreak settings, I’m sorry but I find the BUGG study to be an extremely important and extremely well-designed study with very underwhelming results (negative primary outcome, and a barely statistically significant reduction in MRSA colonization in the context of a substantial mismatch in baseline acquisition rate and (likely) colonization pressure). I agree with you and your co-authors that “replication is warranted”.

  2. Thanks Dan. To be clear, I'm supportive of CP in ICU settings; I agree that outside of ICU settings we may need more data from RCTs. I will say that the ability of ID docs to use antibiotics "off label" without RCT trial data is interesting to compare to our interest in using gowns/gloves off label.

    I'm on the same email list and read the same responses; I would say several were in the "we are cool, we don't do CP category", but we can disagree. There was even boasting about "cojones if you don't use CP." Additionally, most are using CP based on passive surveillance or perhaps active surveillance, which I don't think will be good enough based on my assessment of poor hand hygiene compliance and transmissibility of the organisms (based on contamination of gloves/hands that I presented in the post and we measured in the Morgan ICHE study, which looked at GNR)

    I also think adverse outcome data from non-RCT studies is massively problematic. Since I've been studying adverse outcomes associated with CP for 5+ years, as Dan Morgan's mentor, I've come to the conclusion that the adverse outcome story is overblown. I also think that the BUGG RCT would have identified higher adverse events if they were there, which they weren't. If one still believes the adverse outcomes, the answer is not primarily to ditch CP but develop methods to prevent the adverse outcomes, while preventing the infections.

    Mostly my post was a response to a bias I feel increasing in the field. I feel it is now "not cool" to use contact precautions and you are assumed to be a weak hospital epidemiologist if you use them. This isn't just from comments on this blog or the hospital-epi email list, but from discussions I've had at ICAAC and ICPIC this year. I feel it is important to fight the cognitive biases inherent in this assignment of "coolness" and place them in the context of the current MDRO crisis.

    I disagree on the BUGG study assessment. MRSA colonization pressure was higher in the intervention arm and this would have biased to the null. Even with that, MRSA was reduced. The idea that we should lump VRE and MRSA together and subject them to a combined statistical analysis and bury the MRSA result is not only poor epidemiology (and statistics), it will also prevent others from doing similar trials looking at multiple organisms in the future. This "combined" outcome was not the idea of the authors and it also makes no sense when you think about how independent MRSA and VRE transmission are.

    If you have $6 million, I would be glad to help find someone to replicate the study. Personally, I think no one will pony up another $6 million. Additionally, the data for CP in ICU settings is at least equivalent to CHG in ICU settings since the CHG studies had many flaws, which have been well outlined. Folks seem to have no trouble using CHG.

    I'm cool with being a lone wolf supporting CP in ICU settings; I guess I'm pretty cool with not being cool.

  3. This is a great discussion, I hope some of our other readers will join in (if they don’t, I’ll just continue it over a beer with you at one of our preferred Iowa City watering holes). I agree that CP remain an essential and valuable intervention for MDRO prevention, but I don’t think we’ve been “doing it right”. I could be convinced, as I alluded to in yesterday’s post, that universal ICU CP is a better approach than hospital-wide micro driven CP use as most of us currently practice. I’m just not convinced quite yet.

    You (or one of our readers) can help convince me by answering a couple more questions: one about the MRSA acquisition rate/colonization pressure issue and one about “where we go from here”.

    Regarding the higher baseline rate of MRSA acquisition in the treatment versus control arm of the BUGG study, I am still not convinced that regression to the mean is excluded as an explanation for the MRSA outcome difference—there is a reason to try to match on the acquisition rate, isn’t there? That matching didn’t work as intended, leaving the possibility that the units randomized to the intervention happened to be over-represented by those on a higher part of their 'sine wave' of MRSA ebb and flow during the baseline period, and thus more likely than not to have less extreme rates during the intervention, irrespective of the intervention itself…..in addition, one would expect that a setting of higher colonization pressure would favor effectiveness of CP (i.e. the higher the colonization pressure, the more effective would be universal CP)….and lead to questions about whether the finding is generalizable to settings with lower colonization pressure. Does any of the above make sense, or am I just in desperate need of another cup of coffee?

    And regarding the “where to go from here” question: as someone with a lot of clinical and research activity at the VA, would you advocate that the entire VA system adopt universal ICU gowning and gloving? If you don’t expect better data to come anytime soon, it seems like now is the time to make that decision.

  4. I think colonization pressure is a big deal and higher MRSA admission prevalence would be expected to drive MRSA acquisition, which the UGG policy prevented epidemiologically and statistically. That is, unadjusted higher colonization pressure in the UGG arm would be expected to bias to the null. In fact, this is exactly what the data shows, when you adjust for colonization pressure (which was higher in both the pre-intervention and post-intervention periods in the intervention arm), the effect of UGG was much higher.

    As far as the regression to the mean issue, this is thought not to be an issue when differences appear randomly by chance. These units were not cherry picked as being poorly compliant or for high MRSA rates; this was random since they didn't have the PCR results pack in time for randomization. The matching on pre-intervention outcomes is done to increase power, not to prevent regression to the mean. When you think about the larger than 40% reduction in MRSA acquisition that occurred with UGG after adjusting for colonization pressure, I find that data point convincing. Since we would also expect a similar reduction in MSSA transmission, I'm impressed with the benefits.

    I also just don't think we can expect a magical regression to the mean in 10 ICUs all at the same time right with the intervention; I just don't buy that as a possibility; not by >40%. One ICU in one hospital during an outbreak - sure. These are not equivalent conditions.

    I typically refrain from mentioning VA and recommending VA anything since there are national leaders responsible for making those decisions. If I was in charge of policy in ICU settings, as I was at Maryland, I would recommend universal gowns/gloves (we did this).

    I'm always up for a Sanctuary Pub gathering.

  5. I don’t work for the VA, so….I hereby call on the VA to perform a system-wide pragmatic study of universal gown/glove use in ICUs. Due to the MRSA directive, the VA is the largest system with detailed information about MRSA carriage and infection rates in all their hospitals.

    Until they get that done (after the shutdown is over, of course), you’ve convinced me, Eli. Wasn’t that easy? In all seriousness, I AM convinced that universal CP in ICUs is a better way to apply CP than our current house-wide, micro-driven application. I believe this for many reasons that I will delineate in a future post, but it appeals to my general view that horizontal interventions, applied to all patients at high-risk (rather than only those in whom our micro tests reveal a particular organism) are superior.

  6. Thanks Dan. Since neither of us are practicing hospital epidemiologists the world will remain safe from universal CP. One other benefit of a universal approach is ethical in nature. I would love to hear a discussion by an ethicist, such as our own Lauris Kaldjian, on the ethical implications of isolating an MDRO colonized patient to prevent transmission to another patient who alone benefits. While the colonized patient, who acquired the organism for reasons other than there own, gets visited less often and receives no benefits. A universal application of CP gets around may of these ethical issues, I think. But I'm not an expert in ethics.

  7. Sounds like a symposium talk topic for a future SHEA meeting. And don't forget, I retain my associate hospital epidemiologist title here, so all I need to do now is convince the boss and several ICU directors…we can discuss offline!

  8. Eli, the data you show in the table are for glove compliance, not contact precautions compliance. And if I am hearing you correctly, it's really the gloves that you think are important. However, from the healthcare worker perspective, there's a huge difference between universal gloving and universal gowns/gloves. Do you think the gowns add much?

  9. Good question. It is really glove/gown compliance since the data is from our gown/glove/hand contamination study. Since all HCW wore gowns and gloves, you can't really separate the impact of gowns/gloves on reduced hand contamination. If I had to pick, I would pick gloves over gowns.

  10. So, let me add my 5 cents from non-Iowa connected person, specifically about XDR GNR: I don't think a single intervention is enough. That's why we have bundles...so that we can "attack the enemy [CRE, Acinetobacter] from all flanks"!!! So, what intervention is the most important??? Based on CRE bundles implemented in staggered fashion (just 3 or 4) it would seem that doing active surveillances [for CRE] and cohorting patient/staff based on colonization status might be the most "effective".
    Regarding MRSA, VRE and all other Gram-positives (so last millennium...)...they are all too prevalent in the community and healthcare to be doing the above. Maybe I just love GNR too much to be objective.

  11. I’ve been thinking for a few days about Eli’s post. I have to admit it does make me happy to reduce contact precautions. Too many times I have walked into the room of a critically ill or dying patient to find family members around the bed in plastic gowns, which they have worn for hours. If I can safely do away with that, I am happy indeed. In April 2013, we changed our contact precautions policy in the following ways: (1) we stopped isolating VRE or MRSA patients; (2) we continue to isolate C. difficile patients; (3) we continue to isolate MDR-gram negative patients; to simplify who should be isolated, we defined MDR-gram negative as an organism that is resistant to meropenem (our formulary carbapenem) or susceptible only to meropenem; (4) family members are not required to wear gowns and gloves for patients in contact precautions unless they will be visiting other patients or are directly participating in the care of the isolated patient (e.g., bathing). Our change in policy was very easy to implement. Six months into this, we posted our lowest overall ICU infection rate ever (significantly lower than any rate we’ve seen in 16 years of continuous surveillance), despite the fact that we have incorporated the new CDC VAE surveillance paradigm which is much more sensitive for case detection than the old VAP surveillance. Obviously, it’s too early to know the long-term outcome, but at this point we have not seen an increase in infection rates. For those of you planning to attend the SHEA Spring Conference, there will be a session on successful MRSA control without contact precautions.

    Lastly, I have always argued that contact precautions wouldn’t be necessary if hand hygiene compliance rates were very high and this is borne out in Eli’s table— 90% hand hygiene compliance is essentially as effective as 60% hand hygiene compliance coupled with 90% contact precautions compliance. This tells me that we should keep pushing hand hygiene. It may take a while but I think 90% national hand hygiene compliance is achievable.

  12. I've never much been a fan of n=1. That said, I understand why we practice infection control that way since we haven't really had evidence to support our practices in the past. As we gain evidence for our practice, we will likely have to become an evidence-driven service just like general internists are becoming. While I agree that 90% HH is a great target but probably not good enough to prevent the transmission of completely resistant pathogens such as MDR-Acinetobacter or CRE (see last years NIH outbreak, for example).

    I would also add that the table above assumed perfect effectiveness of hand hygiene and I doubt that is a true representation of what is happening now. It is clear that quantity of product used is too low and HCW are sub-optimally applying the product.

  13. I do want to clarify my n=1 comment. I don't think that you (Mike) are practicing n=1; I think you have synthesized the literature and made your decisions based on local issues such as strategic planning based selection of priority HAI targets. My point was that no one else should just do what you are doing. They need to go through the same process, read the RCTs and time-series studies and decide.

  14. Another point, Eli, is that I think in this argument you and I may have different goals. Your goal is to achieve the lowest possible transmission rate. This assumes that an infection-free hospital stay is the be-all-end-all. I'm not quite as purist. Sometimes I think that infection prevention is not as important as other aspects of patient care. For example, I would be ok with taking a dying patient out of contact precautions even if they had a MDR-gram negative rod. I think that contact precautions carry a high unmeasurable burden for patients and HCWs that is hard to factor into the equation, and so I think our goal should be to minimize the use of contact precautions to the degree that we can.

  15. Here's an update: for this past quarter, hospital-wide (~850-beds) we had 1 device related MRSA infection (a CLABSI), and 2 device-related VRE infections (both CLABSI by current definition, though true source likely gut as both were oncology patients). There were no CAUTIs or VAEs associated with MRSA or VRE. No contact precautions for MRSA or VRE for the last 6 months. Housewide device days for the quarter were: urinary cath 11,807, central line 19,610, ventilator 3,431, total device days 34,848.