General ID: November 2018

MDR bacterial infections are associated with high mortality in India. Okay sure, MDR bacterial pathogens are killing patients everywhere, but the impact of antibiotic resistance is likely to be outsized among low and middle-income countries with limited access to next-generation antibiotics, and this is also where the mortality burden of MDRs has been least studied. The authors conducted a retrospective study involving the outcomes of MDR bacterial infections (not including tuberculosis) in ten hospitals across India over a one-year period. Data obtained included demographics, the susceptibility results of inpatient cultures, and in-hospital mortality. Organisms were classified as MDR and XDR based on European and CDC consensus criteria.

A total 5,103 patient records were included in the analysis. The overall rate of mortality was 13%, and was highest (29%) in patients with infections due to A. baumanii. Mortality due to MDR infections was higher for gram-negative than gram positive bacteria (18% vs 11%), particularly for the subset of patients hospitalized in the ICU (27% vs 16%). In multivariate analysis, mortality was associated with infection with an MDR (OR 1.6) or XDR (OR 2.7) pathogen. This was true for both patients in and outside of the ICU, and again was primarily driven by high mortality associated with resistant gram-negative infections. Curiously, unlike E. coli and K. pneumoniae, infection with MDR and XDR P. aeruginosa were not significantly associated with higher mortality than infection with other pathogens. Patients with infections due to A. baumanii had higher mortality when infected with MDR isolates (OR 2.8), when hospitalized in the ICU (OR 1.7), and when the infection involved the lower respiratory tract (OR not given). For the gram-positive infections, neither methicillin resistance in S. aureus nor glycopeptide resistance in enterococci were associated with mortality.

The authors conclude that in a low-to-middle income country setting, MDR and XDR gram-negative infections are associated with 2-3 fold increases in mortality versus infections with susceptible organisms. 30407501  


Pasteurization of breast milk prevents postnatal CMV infection in very preterm infants. Cytomegalovirus can be transmitted by breastfeeding and can cause severe – sometimes fatal - illness in very preterm infants. At the same time, giving breastmilk to premature infants appears to lower their risk of necrotizing enterocolitis (reference: 29926476). So why not pasteurize the milk?

The authors enrolled infants of CMV IgG-positive mothers from two German NICUs who were <32 weeks gestational age and weighed <1500g  (n=87). The patients were all given breastmilk pasteurized by heating to 62C for 5 seconds from delivery through day four of life. As having a randomized control group for a study like this would be questionably ethical, the authors used a historical control of infants treated at a German NICU 15 years prior who had also undergone testing for CMV. The primary endpoint was CMV status at NICU discharge, assessed by PCR and urine viral culture.

Two (2%) of the infants in the pasteurization study had postnatal CMV transmission, versus 17/83 (21%) of the controls. Adding together each group’s total lifespan spent in the NICU (aka the total time subjects were at risk for CMV infection), the incidence of CMV infection was 0.21 cases/year in the study group versus 1.7 cases/year in the historical group (RR 8.3; p<0.001).

The American College of Pediatricians and CDC already recommend that donor breast milk be pasteurized; this study suggests that for very preterm infants in the NICU, this recommendation should be extended to own mother’s milks as well. I wondered if pasteurization might destroy some of the immunologically beneficial components of breast milks, so I looked it up: fortunately, pasteurization does not affect concentrations of human milk oligosaccharides (which ward off numerous enteric infections) and only modestly affects IgA and IgM (references: 29108433 and 28742378). 30407512


What constitutes adequate empiric antimicrobial therapy for diabetic foot infection these days? The 2012 IDSA guideline for the treatment of diabetic foot infections (DFIs) suggests that moderate to severe DFI be treated with antibiotics active against aerobic gram-positive cocci like S. aureus (i.e. MRSA) and gram-negative bacilli, not necessarily including P. aeruginosa unless specific risk factors are present. This study’s authors examined the microbiology of DFI in patients treated at the Detroit Medical Center between 2012 and 2015. Their goal was to identify clinical predictors of infection with multi-drug resistant organisms (DFI-MDRO; this definition included all the usual acronym pathogens, and I’m not going to mention it again because to me it’s not very interesting) and infections with pathogens resistant to recommended treatment (DFI-PRRT, defined as DFI with an aerobic pathogen resistant to both vancomycin and ceftriaxone – I thought this was the more relevant/practical definition, as it really gets at who might benefit from expanded empiric therapy).

Six hundred forty-eight patients were included; the mean age was 58, and the cohort had a collective 963 episodes of DFI with a total 2450 identified pathogens. Nearly three quarters of infections were polymicrobial; a median two pathogens were recovered in each DFI. The most common pathogens across the entire study period were MSSA/MRSA (present in 57%), Enterobacteriaceae (present in 49%), streptococci (32%), enterococci (31%), and anaerobes (29%); P. aeruginosa was present in only 14%. Most of these organisms were recovered from what you’d consider “legitimate” DFI cultures (e.g. operative cultures of bone or deep tissue), but 22% were bedside swabs. The supplementary materials include a breakdown of which organisms came from which type of cultures, and it looks like the Enterobacteriaceae were overrepresented in the bedside swabs, whereas S. aureus and P. aeruginosa were more likely to come from deep cultures when present.

The incidence of DFI-PRRT was 30%. P.aeruginosa accounted for just under half of the DFI-PRRT organisms, with most of the remainder being VRE and ceftriaxone-resistant Enterobacteriaceae. Patients with DFI-PRRT were more likely to have been admitted from a long-term care facility (14% vs 8%; p=0.006) and to have been hospitalized in the past 90 days (47% versus 23%; p<0.001). In multivariate analysis, predictors of DFI-PRRT included history of colonization or infection with a PRRT (OR 2.5), peripheral vascular disease (OR 2.4), prior use of a beta-lactam / beta-lactamase inhibitor combo (OR 2.2), and CKD (OR 1.6).

So what’s new here? For one, the prevalence of P. aeruginosa was higher than typically reported in DFI (15% vs usually <10%), and both VRE and ceftriaxone-resistant Enterobacteriaceae were also remarkably common. These data suggest that vancomycin and ceftriaxone may no longer be reliable as routine empiric therapy for DFI for except in the absence of specific risk factors for DFI-PRRT. In addition to the usual MDRO risk factors, they also identified prior receipt of BL/BLI combos as a specific risk factor for DFI-PRRT - so there’s one more reason not to use pip-tazo.

While not a criticism per se, I think it’s a shame the authors didn’t calculate likelihood ratios for the identified predictors of DFI-PRRT and come up with a simple clinical risk score to determine who might benefit from empiric expanded antimicrobial therapy. Maybe they’ll do that in a subsequent paper. 30402532


Inadequate perioperative prophylaxis is one of several modifiable risk factors for vascular graft infections. The authors sought to identify predictors of vascular graft infection. To do this, they examined patient outcomes in the prospective Vascular Graft Infection Cohort (VASGRA), a registry at the University Hospital of Zurich, among adults who underwent surgery between 2013 and 2017 and had follow-up at any point within 4.5 years of their initial surgery.

A total of 438 patients, who collectively underwent 768 vascular surgery procedures during the study period, were included in the analysis. The population was 83% male and the median age was 71. Patients who developed graft infection were more likely to have a family history of cardiovascular disease (76% vs 51%), severe peripheral arterial disease (18% vs 3%), and arterial disease involving the lower extremities (49% vs 16%). Procedural factors associated with graft infection included longer surgery (median 4 hours versus 2 hours for those without infection) and lower rates of prophylactic antibiotic administration in the hour preceding surgery (64% vs 79%).

In multivariate analysis, the strongest risk factors for vascular graft infection (p < 0.001 for all) included incisional surgical site infection (aHR 10.1), noninfectious postsurgical complications (aHR 6.5), and prolonged procedure time (aHR 1.2 per each additional hour); however, inadequate perioperative prophylaxis also predictor graft infection (aHR 2.9; 95% CI 1.2-7.1). For the record, the authors’ definition of inadequate perioperative prophylaxis encompassed several criteria, including initiation of prophylaxis less than 24 hours prior to surgery (i.e. before achievement of steady-state drug levels), failure to redose antibiotic prophylaxis during prolonged procedures, and failure to dose antimicrobials appropriately for body weight. So, not only was their definition of inadequate perioperative prophylaxis more than just “we forgot to give the dose of drug,” but it seems they routinely given longer durations of antimicrobial prophylaxis than the typical single perioperative dose recommended in the US. 30395220