March 2019: Antimicrobial stewardship and infection control

We often inflict unnecessary antimicrobials on patients pursuing comfort care.  There are two nice articles out on the same subject this month.  In the first, the authors surveyed patients with advanced cancer transitioning to comfort care who had abnormal urine testing.  Of 300 patients, only 19% had a symptomatic UTI; of the remainder, 21% had asymptomatic bacteriuria or candiduria, and 14% received inappropriate therapy, for a cumulative 279 antimicrobial days. Bacteriuria or candiduria predicted inappropriate antimicrobial prescription: IRR 16.9 for urine with 10k-100k cfu/ml and 27.9 for urine with >100k cfu/ml.  These data match previous studies showing that physicians erroneously assign meaning to the degree of bacteriuria/funguria/pyuria when deciding whether to (inappropriately) treat patients without urinary symptoms. 30821230

In the second study, the authors performed a secondary analysis of a cluster-randomized crossover trial of universal versus physician-directed palliative care consultation at two ICUs at a single medical center.  For this analysis, the outcomes of interest were the proportions of patients who 1) received antimicrobials at discharge, and 2) changed their resuscitation preferences.  The researchers included all adults admitted during weekdays at high risk for morbidity and mortality, as determined by prespecified institutional screening criteria for palliative care consultation.

The study enrolled 242 patients, of whom 199 were eligible for analysis and of whom 132 both 1) were not receiving antimicrobials on admission and 2) survived to discharge.  Among those 132 patients, 27% changed their code status to Do Not Resuscitate / Do Not Intubate, and these patients were less likely to receive antimicrobials at discharge (17% vs 38%; p=0.02), and had shorter total durations of antimicrobial therapy (6.5 vs 9 days; p=0.04).  After discharge, 116 patients (88%) either had 30 days of follow-up data or were known to have died within 30 days, and within this cohort, the authors state there was no difference in antibiotic prescription between those who survived or died.  It would have been nice to have the exact proportion of deaths in each arm, and to have that data stratified by patients who did vs did not have a change in code status, but I won’t complain too much since this is just a short letter to the editor.  Still, these data suggest that perhaps some of the antimicrobials spared in patients who decided to change their code statuses weren’t necessary in the first place. 30838966


Ciprofloxacin is the new tobacco: second-hand exposure to quinolones increases your risk of carrying resistant bacteria.  The authors performed a population-based case-control study examining the association between total neighborhood fluroquinolone consumption and probability of fluoroquinolone-resistant E. coli in urine cultures from patients in said neighborhood over a four-year period.  These data were collected from the largest provider of Israel’s compulsory healthcare system, which serves more than half of the nation’s total population and was able to stratify patient data by neighborhoods.  The researchers excluded neighborhoods with fewer than 60 clients of the healthcare system, all children and young adults below age 21, patients residing in long-term care facilities, and patients likely to receive extended quinolone treatment or prophylaxis (e.g. patients undergoing treatment for cancer, organ transplantation, or receiving immunosuppression).  Data of patients who underwent urine culture were obtained and stratified by one of three culture results: E.coli resistant to quinolones, E.coli susceptible to quinolones, or no growth.  After an individual-level analysis, the authors performed a clustered analysis by patient neighborhood.

The data set encompassed 1733 neighborhoods, 6.9 million people, and 2.4 million patients who had a urine culture with one of the three results listed above.  Ninety percent of episodes of E.coli bacteriuria occurred in women; among them, the quinolone resistance rate was 17% versus 30% for the men.  Patients with quinolone-resistant E.coli were older, more likely to have spent time in a nursing home, and more likely to have been hospitalized in the year before their urine culture.  Unsurprisingly, patients who had previously taken quinolones were more likely to have resistant E.coli in their urine, and this risk increased with greater prior quinolone use in a dose dependent manner.  What’s interesting, though, is that total neighborhood quinolone use was also associated with having a quinolone-resistant E.coli in your urine culture, and that this also showed a clear dose-response relationship. While conferring little additional risk to a single patient, this association affects antibiotic resistance transmission to the degree that the authors calculate that up to 25% of the cases of quinolone-resistance in E.coli from urine cultures could be attributed to “second-hand quinolone exposure.” So here’s one more reason to not prescribe (and to tell your patients not to take) unnecessary antibiotics: in a perverse twist on herd immunity, taking antibiotics may make you an “ecologice niche” through which resistant bacteria can spread into your community.  30846277

Two doses of varicella vaccination provides superior protection to single-dose vaccination in children. This study was a follow-up of a previous multicenter RTC comparing the efficacy of two doses of the combined MMR-varicella vaccine versus a single dose of MMR and single dose of varicella versus two doses of MMR, all 6 weeks apart (or in other words, comparing 2 versus 1 versus 0 doses of varicella vaccine over a 6-week period), for the prevention of varicella disease.  Cases of varicella were defined by DNA testing and/or a clinical + epidemiologic assessment and adjudicated by a blinded independent committee. Symptom severity was assessed by a previously established standardized score.

A total 5803 children were vaccinated as part of the study (2279 in the 2-dose varicella vaccine group, 2266 in the single dose varicella group, and 744 in the MMR + MMR group); the mean age at vaccination was 14 years.  Over a median 9.8 years of followup, varicella occured in 3% of children who received 2 doses of varicella vaccine, 21% of children who received one dose of varicella vaccine, and 47% of children who received no varicella vaccination.   The overall vaccine efficacy of the 2-dose series was 95% versus 67% for the single-dose vaccine.  When restricting the analysis to children who developed moderate or severe varicella, the rates were 0.3%, 3%, and 24% for the two-dose, single-dose, and no varicella vaccine recipients, respectively; by these criteria, the 2-dose vaccine series was 99% efficacious versus 90% for the single vaccination.

Any way you cut it, it looks like two doses of the varicella vaccine confer meaningfully greater protection against varicella than one dose.  These data support current recommendations to vaccinate children with two does of varicella – though given that the median age at the start of vaccination in this series was 14, and the current schedule has the second varicella dose at age 4-6, I wonder if it’s worth studying the effect of a third booster vaccination in adolescence. 30765242

When a patient colonized with C.difficile goes on to develop that infection in the hospital, how often is it due to the colonizing strain?  This has been studied off and on and is a point of contention between proponents of screen-and-isolate (or screen-and-prophylaxis/decolonize) programs to control C.difficile infection (CDI), and those who argue that C. difficile colonization is mostly a marker of morbidity and susceptibility to developing CDI after antibiotic therapy.  The difference between these camps is where you put your infection control money: do you implement C.difficile rectal screening and isolation or treatment, or do you double down on antimicrobial stewardship efforts?

The authors sought to determine the frequency of CDI due to patients’ admission colonizing strains in patients admitted to the Cleveland VA hospital over a six-month period.  Patients were eligible if they had at least two days of hospitalization, did not have diarrhea on admission, and had not had an episode of CDI in the past two months.  Initial identification of C.difficile was done by stool culture; thereafter, CDI was defined as a positive PCR test in a patient with >3 unformed stools in a 24hr period (note this is a generous definition – and so the authors potentially captured patients with some other sort of diarrhea who happened to have a persistently positive nucleic acid test for C.difficile in their stool, which they not may have happened in as many as a third of the cases). The authors established the identities of the colonizing and infecting strains using whole-genome sequencing, and categorized strains as genetically related, possibly related, or unrelated based on the number of differences in single nucleotide polymorphisms.

A total 480 patients enrolled in the study (mean age 67; 95% male; 44% hospitalized in the past year; 63% given antibiotics in the past 3 months).  Of these patients, 68 (14%) had positive C.difficile stool cultures on admission, and 8 of those (12%) were diagnosed with CDI versus 5/412 (1%) of those who did not have positive admission cultures. In the whole genome sequencing analysis, the authors found that 50% of the infecting strains were related to the patient’s colonizing strains. 

So, it seems reasonable to conclude that rectal C.difficile screening protocols could identify, at most, half of the future cases of CDI among patients who are colonized with C.difficile on admission.  Pardon the expression, but in this case, is the juice really worth the squeeze? 30855075