July 2019: Antimicrobial stewardship and infection control

Want to reduce antibiotic use? Improve rates of vaccination. This nice systematic literature review examined RTCs and observational studies assessing the correlation between patients’ receipt of either a vaccine or placebo and subsequent rates of antibiotic use.  The authors identified a total 96 studies on this topic, including 24 randomized-controlled trials.  Most of these related to pneumococcal vaccination (8 RTCs and 50 observational studies) or influenza vaccination (16 RTCs and 14 observational studies) and I’m only going to comment on the data for these two immunizations.

Two RTCs with a total sample size of almost 48,000 patients suggested reduced rates of antibiotic prescription/purchase in children given pneumococcal vaccinations versus placebo (RR 0.9 with 95% CI 0.9-1.0).  A single RTC (n=144) suggested that adults with COPD who received the pneumococcal vaccine subsequently received fewer courses of antibiotics than those given placebo (RR 0.24; 95% CI 0.1-0.5).  On the other hand, two other RTCs (one in adults with COPD and one in children) found no differences in antibiotic prescription between pneumococcal vaccine recipients and non-recipients.

For the influenza vaccine, three RTCs (n=610) supported a reduction in the number of antibiotic courses prescribed to children given vs not given the vaccine (ratio of means 0.62 95% CI 0.5-0.7).  A large and high-quality RTC (n=4253) showed that healthy adults during influenza season who developed a febrile illness were given antibiotics for 28% fewer days if they had received the vaccine (95% CI 16% - 38%), and another large trial in children (n=10,985 person-seasons) showed that the protection from antibiotic prescription extended to the vaccine recipient’s family and community contacts (RR 0.7 with 95% CI 0.6-0.8).  For other measures, the evidence was judged as low or very low certainty – this included one RTC showing reduced antibiotic prescription after vaccination of adults with chronic bronchitis, and nine RTCs showing no effect of influenza vaccination on antibiotic prescription in a variety of settings.  The observational studies are not commented on much in the main text – perhaps because the majority of these were deemed be at high to critical risk of bias.

So, while the evidence is not unanimous, the highest-quality data support a link between higher rates of pneumococcal and influenza vaccination and reduced antibiotic prescription.  Making sure patients are up to date on immunizations isn’t just good primary care, it’s good antibiotic stewardship. 31284031

Documented penicillin allergy in the EMR predicts poorer ID care and poorer infection-related outcomes.  Ask any ID doc/APP/PharmD worth their salt what their favorite antibiotic is and they’ll probably name a beta-lactam (if they say doxycycline, fine – ask for their second favorite).  This class of antibiotic is highly potent, clinical data have demonstrated their superiority in certain scenarios (e.g.; treatment of severe MSSA infections, vs treatment with vancomycin), and alternative agents are often either more toxic (e.g. colistin), dubiously efficacious (e.g. tigecycline), or both.  So, it should be no surprise that patients who have a documented penicillin allergy, in whom clinicians are famously gunshy about prescribing other beta-lactams despite data demonstrating their safety (e.g. 27016799 17321857 26038245), do poorly relative to their peers without a documented allergy.

(I write “documented allergy” rather than “allergy” here because studies show that as many as 90% of patients with a penicillin allergy listed on the chart aren’t actually allergic to penicillin. See: 27888034)

For this study, the authors extracted EHR data from 2.3 million patients at NHS general practice clinics over a one-year period.  They identified patients with a documented pencillin allergy (PenA) and matched them to controls with no penicillin allergy.  They found the prevalence of PenA to be 6%, and to be more common in older people, women, and patients with a medical comorbidity.  After matching, the researchers found that these patients were less likely to receive a penicillin (RR 0.15) and more likely to receive clindamycin (RR 5.5), a macrolide (RR 4.0), a quinolone (RR 2.1), a cephalosporin (RR 2.1), or a tetracycline (RR 1.9);  rates of trimethoprim and nitrofurantoin were also higher, but not to a clinically meaningful degree.  The total number of antibiotic prescriptions received was higher in PenA patients, and specifically they were more likely to receive a second prescription of a new antibiotic class after their initial prescriptions (RR 1.3).  In addition, PenA patients were more likely to develop MRSA colonization or infection (RR 1.9) and were more likely to have died during the study period (RR 1.1).

Take home point: documented penicillin allergy is associated with increased use of second-line antibiotic agents, an overall increase in antibiotic exposure due to initial treatment failure, increased colonization with drug-resistant pathogens, and increased mortality.  So, if you see a patient with a penicillin allergy,  sort it out! 31225607

Rubbing with alcohol santizer for 15sec is as effective as 30sec at eliminating S.aureus and E.coli from the hands.  They should have compared 30 seconds to 3 seconds – the average time it takes to reach the bedside after you wave your hand under the Purell dispenser while striding through the patient’s door – as that’s what I’ve most often observed as actual practice. But so it goes. 

In this study, the researchers enrolled 18 healthcare workers to wash their hands with 60% isopropanol hand sanitizer for either 30 or 15 seconds.  First, the researchers “spiked” each participant’s hands with broth containing 10^6 or 10^8 cfu/ml of either S.aureus or E.coli.  Then, they had the participants touch their fingertips to a culture plate, rub their hands with sanitizer for the indicated period of time, and then touch the plates again.  Their outcome of interest was the reduction in CFU on the plates after vs before the hand-rubbing. They found the median reduction in CFU count was 2.1 log10 cfu and did not vary by the duration of handwashing (between group differences -0.06 log10 cfu, 95% CI -0.34 to 0.22).  The lack of association persisted in an analysis stratified by organism and in all other prespecified subgroups. 31203871

Reduced use of empiric carbapenems for febrile neutropenia in patients with acute leukemia leads to reductions in colonization with VRE.  Must be nice to have an antibiogram with gram-negative agents other than carbapenems that are reliable enough for empiric therapy of febrile neutropenia (FN).  The authors report their center’s experience with changing their empiric drug of choice for FN from a carbapenem to a cycling regimen (either cefepime or pip-tazo, varying monthly).  The outcome of interest, colonization with VRE, was assessed with stool cultures performed on admission and then weekly while the patients remained hospitalized.

A total 214 patients were admitted during the empiric carbapenem period and compared to 128 patients admitted after the change to the cycling empiric regimen; the groups were similar, with median ages of 57-60 and AML as the primary diagnosis in >75% of cases.  While the rates of VRE colonization on admission remained the same throughout the study (2-3%), rates of VRE acquisition in the hospital were markedly reduced after the switch from empiric carbapenems to cycling cefepime and pip-tazo (30% vs 11%; p<0.001).  In a multivariate regression analysis including age, length of stay, and durations of specific antibiotics given, only days of carbapenem exposure was associated with VRE colonization (HR 1.9; 95% CI 1.0-3.4).  There were no differences between patients given carbapenems or cefepime/pip-taz in rates of VRE bacteremia, rates of C.difficile infection, in-hospital mortality, or survival at either 30 days or 1 year after discharge.  On the other hand, patients treated during the carbapenem portion of the study had a 3 day longer length of stay  (30 vs 27 days; p<0.005) and 10% higher total hospitalization costs ($96k vs $87k, p=0.17).

While this study failed to identify an association between greater VRE colonization and increased VRE infections, such associations have been found elsewhere (28480243).  So, forgoing empiric carbapenems for FN – again, if your antibiogram allows it – it may yield downstream benefits. 31046849

Creating a personal protective equipment (PPE)-free zone at the entryway of patient rooms improved (clinician-reported) communication with patients on contact isolation without reducing compliance with PPE or hand hygiene.  There are a few things to discuss in the preamble here.  First, the data supporting contact isolation vary by organism.  Second, many of the trials showing benefit implemented PPE as part of a bundle intervention, which is helpful in validating a specific institution’s policy, but ultimately leaves open the question of whether any specific bundle component is efficacious.  Third, recent published experiences with discontinuing contact isolation policies have suggested that, at least in the context of MRSA and VRE and in centers with high clinician adherence to hand hygiene, contact isolation does little to prevent nosocomial infections or colonizations.  Fourth, data suggest that contact isolation is associated with less frequent visits and communication from doctors to their patients, worsening anxious and depressive symptoms in patients, and lower patient satisfaction.

So: would it help if we created a “PPE safe zone” in the entryway of patient rooms, so that doctors could come for a quick chat without having to don PPE, and would those clinicians remain adherent to other infection control behaviors?

The researchers conducted their study at two academic medical center and one veterans’ hospital; in total, this included nine medical-surgical and specialty wards and seven ICUs.  The units shifted from control to intervention policies (i.e. instituting the PPE-free zone) using a quasiexperimental before-after design.  During the intervention, the study staff created a 1.2 meter zone inside the entry of each patient room, specifically selected to exclude access to high-touch surfaces (e.g. sinks, curtains, cabinets) and to remain at least 2 meters from the head of the bed.  These zones were marked with red duct tape, which were put down and taken up as patients were placed on or taken off isolation in order to avoid habituation.  The study outcomes included adherence to hand hygiene and PPE donning (though the authors don’t mention that this was done covertly, which could have yielded higher adherence via the Hawthorne effect) and clinician’s reported perception of ease of communication and frequency of visits with patients.

The authors observed 1,221 isolation room entries before and 2,749 entries after the switch to a PPE-free zone.  The proportion of clinicians who stayed within the area of the PPE-free zone increased after the intervention, from 4% to 11% (p<0.001).  Compliance with hand hygiene declined over the course of the study, both while units were under the control and intervention conditions; after multivariate analysis adjusting for the facility, unit, and use of gloves, implementation of the PPE-free zone was not associated with reduced hand hygiene (RR 0.9 with 95% CI 0.8-1.1). The authors further write that there were no changes in the rates of appropriate PPE use after the PPE-zone’s introduction, which I find strange given that Table 3 cites  a relative risk of 1.2 with 95% CI 1.0-1.3 and p=0.009 for the intervention units before versus after the intervention.  If anyone can explain this, please DM me on twitter or send me an email. 

Anyway, they also looked at these outcomes stratified by the type of isolation (contact vs enteric), and found that for patients on enteric precautions, hand hygiene compliance dropped over time in the control units but not the intervention units – suggesting that the PPE-free zone served as an important reminder to clinicians to wash their hands after crossing the border to go to the patients’ bedside. Finally, the researchers surveyed the clinicians about the PPE-free zone and found their attitudes were largely positive, with about two-thirds each strongly agreeing that the zones saved time, facilitated communication, and led to more frequent visits with patients on isolation precautions.

My bias that the data supporting contact precautions isn’t all that robust to begin with, and the effect on patient care and satisfaction is real, so if the PPE-free zone can improve clinician-patient interactions without evidence for harm, let’s implement it.  But I’m not an expert in stewardship, so I’d be happy to hear others’ thoughts! 31172904