August 2019: General ID

More than eighty percent of patients with ASB receive antibiotics, for a median 7 days.   This paper out of JAMA IM retrospectively evaluated patients hopsitalized at any of 46 Michigan hospitals over a two-year period, including all those with asymptomatic bacteriuria (ASB - defined as a positive urine culture without documented signs or symptoms attributable to UTI) and compared the patients who did vs did not receive at least one dose of antibiotic treatment.  Specifically, “documented signs and symptoms of UTI” meant urinary symptoms, fever, and dysautonomia or spasticity in patients with spinal cord injury; altered mental status was not considered a symptom of UTI unless the patient had a leukocytosis, hypotension (SBP <90mmHg), or met SIRS criteria.  Outcomes of interest included 30-day mortality, readmission, ED visit rates, length of stay after urine testing, and the incidence of C.difficile infection.

The authors identified a total 2733 patients with ASB; the mean age was 77, 78% were woman, 14% were catheterized, 42% had chronic kidney disease, 39% had diabetes, and 21% had dementia.  Almost all of these patients (94%) had accompanying pyuria, and of the reasons a urine culture was sent (documented in only 57% of cases), abnormal urinanalysis was the most commonly cited reason.  In total, 83% of patients were treated with an antibiotic, for a median 7 days.  In multivariable analysis, factors associated with ASB treatment included older age (OR 1.1 per 10-year increase; 95% CI 1.0-1.2), abnormal urinalysis (OR 2.8; 95% CI 2.1-3.9), >100k cfu/ml bacteriuria (OR 2.3; 95% CI 1.8-2.9), peripheral leukocytosis (OR 1.5; 95% CI 1.2-2.0), and E.coli bacteriuria (OR 1.4; 95% CI 1.1-1.8).  Patients treated vs not treated for ASB had no differences in their rates of 30-day mortality, readmission, ED visit, or CDI; however, patients treated for ASB had a longer length of stay (median 4 vs 3 days; RR 1.4; 95% CI 1.3-1.5). 

So, treating your patient’s ASB won’t reduce their risk of coming back to the ED after discharge, being readmitted, or acutely dying, but it will prolong their hospitalization.  Great!  31449295

When you have a patient with Enterobacteriaceae bacteremia and decide to step down to oral therapy, should you reach for a quinolone, TMP/SMX, or a beta-lactam?  The authors performed a systematic review and meta-analysis of IDWeek abstracts and published studies that reported either all-cause mortality or infection recurrence over a follow-up period of 3 weeks to 3 months.  They identified eight studies (total n=2289), all retrospective.  The most common choice for oral stepdown was a quinolone (65%), followed by a beta-lactam (27%), followed by TMP/SMX (8%). 

All-cause mortality was similar among patients who received an oral beta-lactam versus either a quinolone or TMP/SMX (RR 1.1 with 95% CI 0.7-1.9); however, receipt of a beta-lactam rather than a quinolone was associated with a higher rate of recurrent infection (OR 2.0 with 95% CI 1.2-3.6) and recurrent bacteremia (OR 2.2 with 95% CI 0.9-5.0).  Small sample size limited the value of comparisons with TMP/SMX alone; no differences were observed versus beta-lactams in rates of recurrent infection (OR 1.0 with 95% CI 0.3-3.1) or recurrent bacteremia (OR 0.5 with 95% CI 0.2-1.9), though of course the latter odds ratio would be clinically important if they represented a true difference. Finally, infection-related readmissions occurred almost twice as often among patients transitioned to beta-lactams versus a quinolone or TMP/SMX (OR 1.8 with 95% CI 0.9-3.9).

So, as anxious as the fluoroquinolones make everyone these days, these data suggest they’re the superior choice for oral stepdown therapy in patients with Enterobacteriaceae infections. 31412127

Speaking of oral quinolones and TMP/SMX for serious Enterobacteriaceae infections, here’s a nice retrospective study on the outcome of combination quinolone-TMP/SMX therapy for bone and joint infections due to Enterobacter cloacae.  The authors used this combination in a total of 30 such patients over a 7-year period; all but two cases involved hardware, and all were treated for 8-12 weeks.  The cure rate was 80% (compare this to the average 70% cure rate across all studies of IV or PO therapy for osteomyelitis) with a mean 29 months of followup.  Small N, but until we have a larger study showing IV carbapenems can offer similar outcomes, I’ll contend oral quinolone-based therapy for AmpC-producing gram-negative osteo is the way to go. 31096009

Four versus six weeks of antibiotics for osteoarticular infections after orthopedic hardware removal yielded similar clinical outcomes in this randomized controlled trial.  The authors of this single-center trial out of Switzerland randomized adults with a first episode of osteoarticular hardware infection (including PJI and infection of plates, nails, and other material) to receive either four or six weeks of antibiotics after hardware removal.  They excluded patients who had immediate implantation of new hardware (but not later implantations, i.e. 2-stage exchange of PJI), as well as those with incomplete source control, other retained foreign material at the site of infection, infection due to unusual organisms (e.g. fungi, mycobacteria), or endocarditis.  They defined infection as positive operative cultures in a patient with compatible clinical symptoms.  The primary outcome was “remission,” defined as the complete absence of clinical, lab, or radiologic findings suggesting infection after at least 6 months of followup.

The intention to treat analysis included 122 patients (there seems to be a discrepancy in the manuscript here – figure 1 states 123 patients, while the text states 123 infections in 122 patients.  I’m going with the text); the mean age was 64, 60% were men, and a third were “immunocompromised” on account of diabetes, active cancer, immunosuppressive medications, dialysis, or cirrhosis.  The two arms of the study were well-balanced. S. aureus was the most common single pathogen, seen in 37% of cases, and all but four of these were MSSA.  The mean duration of antibiotic therapy in the 4 and 6 weeks arms were 28 days (range 27-30) and 42 days (range 39-45); in both groups, the duration of initial parenteral therapy was 4 days, and the most commonly selected agents for definitive therapy were (oral) quinolones, clindamycin, doxycycline, TMP/SMX, and amox/clav and (IV) vancomycin, amox/clav, and cefuroxime. Surgical management was similar between groups, with a mean 1.5 vs 2 debridements in the 4 vs 6-week arms.

116/123 cases of infection (94%) achieved clinical cure over a mean 2.2 years of followup.  The primary outcome of remission was achieved in 58/62 cases in the 4 week arm vs 58/61 cases in the 6 week arm (difference in remission rate 1.5% with 95% CI -9.8% to 6.8%).  Similar results were observed in the per-protocol analysis (risk difference 2% with 95% CI -8.4% to 7.9%).  In total, 19% of patients had a potentially antibiotic-associated adverse event, and the incidence of this was similar between the 4 and 6-week arms (risk difference 5% with 95% -12% to +22%).

This is a study that I wish had been continued for another couple of years to boost the sample size and narrow the confidence intervals.  It sure looks like 4 weeks of antibiotics work just as well as 6 weeks after orthopedic hardware removal – the relative rates of remission are nearly identical.  Unfortunately, they chose and powered their study to detect a 10% non-inferiority margin, which I think is probably too big to be acceptable for a lot of clinicians.  So, do I think 4 weeks of antibiotic therapy works just as well as 6 weeks in this situation?  You bet.  But can I fault my orthopedic surgery colleague who doesn’t find these data compelling?  Not really. 31106353

SGLT-2 inhibitors do not appear to increase the risk for severe bacterial UTI.  You might expect to see more UTIs with a drug that dumps sugar into the bladder and thereby turns the urine into a more hospitable culture media.  And indeed, SGLT-2i use has been associated with an increase in candida genitourinary infections and a few reports of necrotizing soft tissue infections (i.e. Fourier’s gangrene).  But, for some reason, bacterial UTI rates seem unaffected.

In this study, the authors reviewed 2 large national databases for patients with diabetes starting either an SGLT-2i or a dipeptidyl-peptidase-4 inhibitor (DPP-4), comparing the two groups for subsequent hospitalization for UTI, sepsis associated with UTI, or episode of pyelonephritis.  >100k patients were enrolled into each of two cohorts based on the two databases. After propensity score matching, the SGLT-2i recipients had similar rates of severe UTI as DPP-4 recipients in cohort 1 (HR 0.98 with 95% CI 0.7-1.4), and actually had fewer such events in cohort 2 (HR 0.72 with 95% CI 0.53-0.99).  These findings were robust across a variety of subgroup analyses, and a second analysis looking at rates of any UTI also found no uptick with SGLT-2i use. 31357213

Here's another study showing bacteremia can be treated with short-course therapy.  The big study on this was last-year’s RTC by Yahav et al, but (1) the more the merrier and (2) this study was not restricted to gram-negative (or for that matter, monomicrobial) bacteremias.  The authors retrospectively reviewed patients who received blood cultures in their ED in Taiwan over an 8-year period, enrolling patients in the cohort if (i) the blood cultures were positive (ii) the patient was hospitalized and treated for 5-16 days (iii) the patient received appropriate empiric and definitive therapy (defined as a drug-bug pair in the Sanford guide to which the isolated organism is susceptible), and (iv) they survived to the end of the course of treatment and (if known to have died) had a certain date of death.  Cases of single blood culture positivity for an organism suspected to be a contaminant were excluded.  The authors stratified these patients into those who received 5-10 days of antibiotics versus those who received 11-16 days. The primary outcome of interest was mortality at 30 days from onset of the bacteremia; 30-day mortality from completion of treatment was a secondary outcome.

 A total 1431 patients were included in the cohort, of whom 71% were treated with short-course therapy and 29% with long-course therapy.  The mean age was 68, half were women, 38% had diabetes, 27% had cancer, and 17% had chronic renal diseases.  With regards to microbiology, 71% of infections were due to Enterobacteriaceae, and only 8% were due to S.aureus.  The average length of stay was 11 days, and the crude mortality was 1.2%.  Overall, the folks in the long-course group were sicker – with higher Pitt bacteremia scores, rates of nursing home resident (7% vs 3%), rates of inadequate source control (5% vs 1%), and more “rapidly or ultimately fatal” comorbidities (25% vs 20%).  To deal with this, the authors generated a 2:1 propensity score matched cohort (total n=1089).  Within this cohort, they observed no difference in the 30-day mortality rate between patients given short or long-course therapy (1% vs 1.9%; p=0.25), a finding similar to that of the unmatched/total cohort.  However, when looking at the days from end of treatment definition of 30-day mortality, short-course therapy was actually associated with a lower mortality rate (1.4% vs 3.6%; p=0.009) and lower rates of subsequent infection (2.2% vs 6.1%) and subsequent infection with an antibiotic resistant pathogen (1.7% vs 4.4%).

This study was still mostly in gram-negatives, and in particular I don’t want anyone to think that this paper is adequate evidence that 5-10 days is enough for S.aureus bacteremia (it’s not).  That said, this more evidence that GNR bacteremia does not require prolonged therapy. 31108223

There are a subset of pleural empyemas with a polymicrobial etiology similar to dental and brain abscesses, suggesting a subset of “primary empyemas” that are of odontogenic origin.  I don’t really have anything useful or practice-changing to say about this article, but include it because it’s very cool to see new diagnostic technologies (in this case, massive parallel sequencing) improving our understanding of the basic pathophysiology of infection. 30580031

Are phages included in the term “microbiota”? If so, does an oral+rectal phage cocktail count as an FMT? Well, anyway, it cleared a patient colonized with a CRE K.pnuemoniae in this case report out of Italy.  This raises an interesting question.  There are several studies suggesting that fecal microbiota transplantation can eradicate enteric colonization by multidrug resistant pathogens.  I’d always assume that, if true, the mechanism would be competition for ecologic niches by “more fit” normal gut bacteria not burdened by the metabolic cost of producing beta-lactamases and relying on mutated ribosomal proteins.  But, perhaps that’s not it at all – perhaps said eradication is accomplished by phages from the donor “virome”? 31414123