There is a sentence sitting in the CDC guideline that governs how American hospitals handle resistant organisms, and it is the kind of sentence institutions write once and hope will fade into obscurity. Describing its own recommendation that hospitals put gowns and gloves on everyone entering the room of a patient carrying a multidrug-resistant organism, the 2006 guideline says: “This recommendation was based on general consensus and was not necessarily evidence-based.” The next sentence concedes that no study had ever compared standard precautions alone against standard precautions plus contact precautions.
That was the state of the evidence when the recommendation was written. It is not the state of the evidence now. The comparison that didn’t exist has since been run four times, randomized, across nearly 150,000 patients. Four times the answer came back: no measurable benefit. The gowns are still going on, converting oil into landfill after a brief stop at the patient’s doorway.
The trials came back
Contact precautions are a mid-1990s outbreak tool that was generalized to routine care without ever being tested in routine care. When the tests finally arrived, they were large and they were consistent.
The cleanest is a 2020 cluster-randomized crossover trial in Lancet Infectious Diseases: 20 non-critical-care wards at university hospitals in Germany, the Netherlands, Spain, and Switzerland, 38,357 patients, 11,368 of them screened at least twice, testing contact isolation for ESBL-producing Enterobacterales, the resistant gram-negatives that trigger gowns in most American hospitals. Acquisition ran 6.0 per 1,000 patient-days with isolation and 6.1 without. Incidence rate ratio 0.99. The authors’ conclusion was that contact isolation “showed no benefit.”
The pattern holds wherever it has been tested. STAR*ICU randomized 18 ICUs to surveillance cultures plus expanded barrier precautions and found nothing. The BUGG trial put universal gowns and gloves on 26,180 ICU patients and found nothing on its primary outcome. And REDUCE MRSA, 74 ICUs and 74,256 patients, is the most damning of the four, because it did not merely show that screening and isolating carriers fails. It showed that ignoring who was colonized and decolonizing everyone beat the isolation strategy on bloodstream infections from any pathogen. Universal decolonization prevented one bloodstream infection for every 54 patients treated. The alternative to the ritual was not chaos. It was a better result.
One asterisk, included because an informed reader will bring it up anyway: in BUGG, MRSA acquisition alone, one component of the null composite, reached nominal significance at p = 0.046, a secondary outcome with no adjustment for multiple comparisons. It is the only randomized transmission signal favoring barriers anywhere in this literature. The authors said it required replication before drawing conclusions. Thirteen years later, none exists.
Ask two countries about the same organism
If contact precautions for ESBL were driven by evidence, practice would look the same wherever the evidence had been read. Here is what practice looks like.
Germany does not recommend screening or isolation for ESBL at all, unless the strain also resists ciprofloxacin. The Netherlands requires special hygiene measures for every ESBL patient. These two countries share a border, a scientific literature, and several of the wards in the randomized trial above. The organism does not change at customs. The policy does.
The United States runs the same experiment internally. Surveys find more than 90 percent of hospitals gowning for MRSA and VRE while more than 30 hospitals do not, and the physicians surveyed report that practices for getting a patient off isolation “varied widely,” with free-text responses describing frustration at how precautions are started and stopped. Inside the VA, the country’s largest health system, a 2026 analysis of 121 hospitals had to survey each facility separately on whether it isolated colonized patients and whether it isolated infected patients, because hospitals split on that line. Many facilities suspended practices during the pandemic. After adjusting for baseline burden, pandemic era discontinuation was not consistently associated with more infections.
Some hospitals have retreated to a middle position: no gowns for colonization, gowns for active infection. It feels like rigor. It has the same evidence base as the position it replaced, a nonexistent evidence base. No completed trial has tested precautions scoped to infection; the trials that failed tested colonized and infected patients together, and the one study now testing infection-only scoping, in French geriatric wards, began enrolling only in 2024. And the contamination data show which variable actually matters. Rooms of infected and colonized patients were contaminated at nearly identical rates, 73 versus 69 percent; the organism does not know whether it is causing symptoms. What predicted contamination was the body site: with MRSA in a wound or urine, 36 percent of surfaces grew the organism, against 6 percent for every other site. The thing worth gowning for is uncontained drainage, not the word infection in a problem list. The academic centers that removed precautions for infected patients with contained body fluids watched their infection rates do nothing.
To be precise about that landing spot: the containment rule is a mechanism call, not a trial result. Contamination is a surrogate, and nobody has shown that gowning for the heavy shedder improves outcomes either. The difference is epistemic. Containment-based precautions are untested and biologically coherent. The endemic default is tested and null. One is a defensible place to wait for evidence. The other is a verdict being ignored.
Practice this divergent is not tracking evidence. The evidence is identical everywhere. What varies is the local default, and the default is doing all the work.
What the ritual costs
The best-documented cost is the one that should bother physicians most: the gown reduces how often anyone is with the patient.
The finding has replicated for two decades. Attending physicians examined 35 percent of isolated patients on morning rounds versus 73 percent of everyone else. Across 7,743 observed visits, isolated patients received 36 percent fewer visits and 18 percent less contact time. And inside the BUGG randomized trial, room entries fell from 5.24 to 4.28 per hour, which means the effect survives randomization. Isolated patients are not different people. The barrier works the way barriers work, on clinicians. CDC’s own outdated guideline concedes the point, noting that personnel, attending physicians included, “were half as likely to enter the rooms of, or examine, patients on Contact Precautions,” citing findings that such patients had more preventable adverse events and less documented care, and then instructing hospitals to “counteract these potential adverse effects.” It does not say how.
The money is simpler. Full staff-time accounting puts an isolation-day at roughly $150, about three-quarters of it labor: six minutes of donning and doffing for every hour a patient is on precautions. Blocked and closed beds cost about $9,400 in lost charges per 100 bed-days. And the intervention being paid for barely exists at the bedside. Where full compliance with the bundle has been measured, it ran 7 percent at one hospital and 22 percent at another, both measured during C. difficile isolation, and removing the gown contaminated the wearer’s skin or clothing in 46 percent of observed removals. The policy being defended is not even close to the policy being delivered.
The gowns do have a defense, and it should be stated outright. Precautions measurably improve hand hygiene at room exit. And the claim that isolation itself causes preventable patient harm, an argument advocates of de-escalation are fond of leading with, is weaker than its reputation: the noninfectious adverse-event literature taken whole is a wash, and the best-designed study, nested inside the BUGG trial, found no excess adverse events under universal gowning. Patient satisfaction surveys show no difference either way. The case against gowns does not need those harms. An intervention that costs $150 a day, keeps clinicians out of the room, and delivers no measurable benefit is indictment enough.
Why nobody takes the gowns off
A practice with this record should be shrinking. It is not, because nothing in the system is built to shrink it.
Start with who owns each outcome. If a hospital stops isolating for ESBL and an infection follows, the decision has a name attached and a paper trail, and the infection gets on the agenda at three different meetings. If the hospital keeps isolating, the costs land as six minutes here, a blocked bed there, a doctor who asks questions from the doorway instead of sitting down, a pump alarm that rings for an extra minute while the gown goes on, a grandchild who keeps her distance because of the sign, a hand held through nitrile when it is held at all. None of it is attributed to anyone. Stopping is a decision. Continuing is a default.
Then look at what gets measured. Gown compliance is auditable: a person with a clipboard can count it, and surveyors do. The benefit is not auditable even in principle. A prevented transmission is an event that never happens; no database anywhere records the infection a gown stopped. So the institution audits the input and treats it as the outcome, gown compliance standing in for infections prevented, with no mechanism to detect when the conversion rate between the two is zero. For endemic ESBL, the trials put it at or near zero. The ritual becomes a metric because it can be measured, and the metric becomes the standard.
Then try to get a patient off isolation. Starting precautions is automatic: a lab result or diagnosis flags, a sign goes on the door. Stopping requires a physician to raise it, infection prevention to agree, and rarely a health department to weigh in, and a misunderstanding at any node results in the sign staying up. The surveys agree: everyone starts precautions the same way, nobody stops them the same way, and the free-text responses keep landing on the word frustration.
The ratchet turns one way at the system level too. In 2007, at the peak of MRSA alarm, Illinois wrote screening-and-isolation into state law, the first state to do so, over the published objection of the two professional societies whose members would have to run it. Across five years of surveys that followed, ICU MRSA colonization prevalence in Chicago’s hospitals sat at 11 percent and never moved. The statute is still on the books. England ran the honest version of the experiment: it mandated universal screening nationally beginning in 2009, audited the program, found 1.5 percent admission prevalence and poor compliance, and replaced it with targeted screening in 2014. Adoption was fast, legislative, and evidence-free. Withdrawal, where it has happened, requires a government willing to audit its own mandate. No American legislature has done that. The retreat here runs hospital by hospital, quietly, and only where a statute written before the trials does not force the practice. Where one does, the evidence is beside the point; the gowns are the law.
And the guideline that would settle the question is frozen. The governing documents date to 2006 and 2007. Every randomized trial above was published in 2011 or later, and for years afterward no revision even began. When a revision finally began in 2022, fifteen years after the guideline it was meant to replace, it cycled through drafts and send-backs without ever producing a final document, and in spring 2025 the committee responsible, HICPAC, was disbanded with nothing delivered. It has not been reconstituted. The recommendation that described itself as consensus rather than evidence now has no body left to revise it, which makes it zombie policy in the strict sense: the body that would revise it is gone, nobody can kill it, and it keeps moving anyway.
Flip the burden
What makes the persistence indefensible rather than merely inertial is that the fine print already agrees. CDC recommends the gowns, but recommending is all it does: routine contact precautions for MDROs carry Category IB, strongly recommended, and pointedly not Category IC, the label the guideline applies where a regulator, OSHA, CMS, or a state, has already made the practice law. CDC does not control that regulation; the IC label only records where it exists. Its absence here records that when the recommendation was written, there was no mandate to record; the state statutes came later, and the guideline was never revised to notice them. As for why the recommendation is strong at all: the grade was assigned in 2006, on observational studies and theoretical rationale, by the same document that admitted the direct comparison had never been run. The strength predates the evidence, and the evidence, when it came, said no. On when to stop, the guideline states that no recommendation can be made. Medicare’s Conditions of Participation require hospitals to run an infection prevention program, not to gown for any named organism, so the binding rule at most hospitals is the hospital’s own policy. And the one place the guideline awards its strongest category is outbreaks, which at least matches the tool’s original purpose: outbreaks are where it earned its reputation, and nobody has run the trial that would unseat it there. The same logic covers C. difficile, carbapenem-resistant organisms, Candida auris, and the patient with draining wounds or uncontained secretions. To be clear about what that list is: it is not an endorsement. I am not advocating for gowns anywhere until they produce outcome data. Continuing them in those settings is reasonable while the question is open, and reasonable under uncertainty is the most they have earned.
What the list does not include is a separate carve-out for every unit that has not hosted its own trial. The ICU nulls already exist for MRSA and VRE; STAR*ICU and BUGG were ICU trials. And the mechanisms that sink the practice, the undetected reservoir, the 7-to-22-percent delivered fidelity, the doffing that contaminates half the time, do not change at the door of a pediatric ward. Demanding a fresh randomized trial for every setting and every organism before the default can even be questioned is not rigor. It is a moat. The first non-inferiority study in geriatric wards began enrolling in 2024, eighteen years after the 2006 guideline, and will not report until 2027; at one subgroup per decade, the moat will hold for decades.
Everywhere else, the question should be the one we ask of every other intervention. Propose contact precautions for endemic ESBL today, as a new practice, carrying four null randomized trials, a $150 daily unit cost, a measured reduction in physician visits, and 7 percent delivered fidelity, and no committee would approve them. They persist because they are the incumbent, and nobody audits the incumbent.
The flip is not “prove isolation is harmful before you may remove it.” The flip is that an intervention with real costs holds its place by demonstrating benefit, on a clock, the way everything else we do to patients must. Isolation during an outbreak can at least point to the containments it was part of. Isolation as a permanent endemic default has had thirty years and four randomized trials to prove itself, and failed.
A drug with four null trials and a daily price tag would come off the formulary. In infection prevention it gets a compliance dashboard. Standard precautions won the trial. The gowns won the policy.
How this is made: the argument, sources, and verification are mine; AI helps draft and edit, and I stand behind every sentence. Full process on the About page.
References
CDC. Management of Multidrug-Resistant Organisms in Healthcare Settings (2006), Prevention and Control. www.cdc.gov/infection-control/hcp/mdro-management/prevention-control.html
Maechler F, et al. Contact isolation versus standard precautions to decrease acquisition of ESBL-producing Enterobacterales in non-critical care wards: a cluster-randomised crossover trial. Lancet Infect Dis 2020. doi.org/10.1016/S1473-3099(19)30626-7
Huskins WC, et al. Intervention to reduce transmission of resistant bacteria in intensive care (STAR*ICU). N Engl J Med 2011. doi.org/10.1056/NEJMoa1000373
Harris AD, et al. Universal glove and gown use and acquisition of antibiotic-resistant bacteria in the ICU (BUGG). JAMA 2013. doi.org/10.1001/jama.2013.277815
Huang SS, et al. Targeted versus universal decolonization to prevent ICU infection (REDUCE MRSA). N Engl J Med 2013. doi.org/10.1056/NEJMoa1207290
Müller J, et al. Cross-border comparison of the Dutch and German guidelines on multidrug-resistant Gram-negative microorganisms. Antimicrob Resist Infect Control 2015. doi.org/10.1186/s13756-015-0047-6
Cimen C, et al. Infection prevention and control without borders: comparison of guidelines on multidrug-resistant organisms in the northern Dutch-German cross-border region. Antimicrob Resist Infect Control 2025. doi.org/10.1186/s13756-025-01528-3
Morgan DJ, et al. Reconsidering contact precautions for endemic MRSA and VRE. Infect Control Hosp Epidemiol 2015. doi.org/10.1017/ice.2015.156
Russell D, et al. Routine use of contact precautions for MRSA and VRE: which way is the pendulum swinging? Infect Control Hosp Epidemiol 2015. doi.org/10.1017/ice.2015.246
Khader K, et al. Estimating changes in facility MRSA infection rates due to changes in MRSA precaution policy. Clin Infect Dis 2026. doi.org/10.1093/cid/ciag176
Saint S, et al. Do physicians examine patients in contact isolation less frequently? Am J Infect Control 2003. doi.org/10.1016/s0196-6553(02)48250-8
Morgan DJ, et al. The effect of contact precautions on healthcare worker activity in acute care hospitals. Infect Control Hosp Epidemiol 2013. doi.org/10.1086/668775
Sharma A, et al. The cost of contact precautions: a systematic analysis. Can J Infect Control 2020. doi.org/10.36584/cjic.2020.004.03
Schrank GM, et al. The discontinuation of contact precautions for MRSA and VRE: impact upon patient adverse events and hospital operations. BMJ Qual Saf 2019. doi.org/10.1136/bmjqs-2018-008926
Yanke E, Zellmer C, Van Hoof S, Moriarty H, Carayon P, Safdar N. Understanding the current state of infection prevention to prevent Clostridium difficile infection: a human factors and systems engineering approach. Am J Infect Control 2015;43(3):241-247. doi.org/10.1016/j.ajic.2014.11.026
Tomas ME, et al. Contamination of health care personnel during removal of personal protective equipment. JAMA Intern Med 2015. doi.org/10.1001/jamainternmed.2015.4535
Croft LD, et al. The effect of universal glove and gown use on adverse events in intensive care unit patients. Clin Infect Dis 2015. doi.org/10.1093/cid/civ315
CDC. Management of Multidrug-Resistant Organisms in Healthcare Settings (2006), Summary of Recommendations. www.cdc.gov/infection-control/hcp/mdro-management/summary-recommendations.html
42 CFR 482.42, Condition of participation: infection prevention and control and antibiotic stewardship programs. www.ecfr.gov/current/title-42/chapter-IV/subchapter-G/part-482/subpart-C/section-482.42
SHEA. Importance of reinstating CDC’s Healthcare Infection Control Practices Advisory Committee (HICPAC). shea-online.org/importance-of-reinstating-cdcs-healthcare-infection-control-practices-advisory-committee-hicpac
Ezzi O, et al. Discontinuation of contact precautions for ESBL-producing Enterobacterales in geriatric acute-care units (GERSP): study protocol. PLoS One 2026. doi.org/10.1371/journal.pone.0353783
Haessler S, et al. Stopping the routine use of contact precautions for management of MRSA and VRE at three academic medical centers: an interrupted time series analysis. Am J Infect Control 2020. doi.org/10.1016/j.ajic.2020.06.219
Edmond MB, et al. The impact of discontinuing contact precautions for VRE and MRSA on device-associated infections. Infect Control Hosp Epidemiol 2015. doi.org/10.1017/ice.2015.99
Boyce JM, et al. Environmental contamination due to methicillin-resistant Staphylococcus aureus: possible infection control implications. Infect Control Hosp Epidemiol 1997;18(9):622-627. PMID 9309433.
Weber SG, et al. Legislative mandates for use of active surveillance cultures to screen for MRSA and VRE: position statement from the Joint SHEA and APIC Task Force. Am J Infect Control 2007. doi.org/10.1016/j.ajic.2007.01.001
Lin MY, et al. Regional epidemiology of methicillin-resistant Staphylococcus aureus among adult intensive care unit patients following state-mandated active surveillance. Clin Infect Dis 2018. doi.org/10.1093/cid/cix1056
MRSA Screening and Reporting Act, 210 ILCS 83, Illinois Compiled Statutes (enacted by Public Act 95-312, 2007). www.ilga.gov/Legislation/ILCS/Articles?ActID=2919&ChapterID=21
Department of Health (England). Implementation of modified admission MRSA screening guidance for NHS (August 2014). assets.publishing.service.gov.uk/media/5a7ed4f1ed915d74e6226bc7/Implementation_of_modified_admission_MRSA_screening_guidance_for_NHS.pdf
CDC. Update on the Part One draft update to the Guideline for Isolation Precautions. Safe Healthcare Blog. blogs.cdc.gov/safehealthcare/draft-2024-guideline-to-prevent-transmission-of-pathogens-in-healthcare-settings