April 2019: ID Diagnostics

Who develops sepsis after urinary stone removal, and what are the value of stent and stone cultures?  The authors prospectively studied patients undergoing percutaneous nephrolithotomy or uteroscopic stone removal between 2005 and 2016, collecting data om patient demographics and comorbidities, previous UTIs, and postoperative course.  They compared patients who did versus did not meet SIRS criteria in the 48 hours after their procedures; the authors additionally compared the concordance of the patient’s preprocedural midstream urine cultures and subsequent urinary stent or stone cultures.

A total 2,116 patients underwent stone removal, of whom 1011 had both a urine and foreign body culture (that is, urinary stone or stent culture) and were included in the study.  The mean age was 53, two-thirds were men, and a fifth had had a UTI in year prior to the procedure.  Postoperative SIRS occurred in 7% of patients, of whom 47% had a positive preoperative urine culture and 74% had a positive foreign body culture.  In multivariate analysis, only positive foreign body culture (OR 10.3; p<0.001) and percutaneous versus ureteroscopy nephrolithotomy (OR 2.0; p=0.02) were associated with developing SIRS.  The microbiology of the foreign body cultures was more diverse than the midstream urine cultures, with more frequent isolation of E. faecalis, S. aureus, CONS, and viridans strep (collectively, 39% versus 18%) and less frequent isolation of E.coli, K.pneumoniae, and P.mirabilis (collectively, 34% versus 60%).  Among the patients who became bacteremic, the blood culture more often matched the foreign body culture (60%) than the preprocedural midstream urine (37%).

Take-home message?  In nephrolithotomy, preprocedural urine cultures predict neither postoperative sepsis nor the causative organism in cases where sepsis occurs; foreign body cultures, on the other hand, correlate pretty well with both.  30972587

Speaking of urinary tract infections, here’s another paper asking whether you need to get follow-up blood cultures for bacteremic UTI.  The authors retrospectively examined 306 patients with bacteremic UTI who had followup blood culture.  Of these, 55 (18%) had a positive follow-up culture, and this was associated with diabetes, cancer, complicated UTI, and initial ICU admission. The patients with positive follow-up cultures had longer times to defervescence, but no greater rate of subsequent bacteremia.  So, a positive followup blood culture may be a marker of morbidity or infection severity (rather than necessarily indicating an uncontrolled source of infection – though it certainly can do that too), but it’s not clear that having a positive followup culture affects the overall outcome of the infection. 30689071

In elderly patients with sepsis, low serum albumin and inadequate empiric antimicrobials predict death.  I had an attending in residency who once called serum albumin the “gonna-die-o-meter.” For infectious in elderly patients, at least, that seems to be an apt description. This study was a retrospective analysis of elderly patients with sepsis admitted through the ER to a single teaching hospital over a one-year period.  The authors aimed to describe the prognoses of these patients and the prognoses’ determinants. A total 235 patient-episodes occurred during the study period, among which 22% met criteria for severe sepsis or septic shock.  Of these, 72% were community-acquired infections (vs nosocomial or healthcare-related), and the three most common sites of infection were urinary (54%), respiratory (26%), and intraabdominal (12%). 

Empiric antimicrobial therapy was adequate (defined at least one antimicrobial active against the isolated pathogen having been administered at the recommended dose) in 83% of episodes.  Survivors more often received adequate empiric therapy vs non-survivors (14% vs 34%; p=0.013).  All-cause mortality was 5% at two weeks and 12% and four weeks.  In multivariate logistic regression, factors associated with 30-day mortality included albumin < 2.6 g/dl (OR 3.26; p=0.029), Charlson score (OR 1.23; p=0.012), and inadequate empiric antimicrobial therapy (OR 3.3; p=0.039).

If I have one critique of this study, it’s that they chose to analyze serum albumin as a categorical rather than continuous variable (i.e. splitting the cohort into those with albumin values <2.6 vs >= 2.6 g/dl rather than assessing the risk of death along the continuum of albumin values).  Given that albumin tracks with severity of infection as well as host status, I suspect that someone coming into the ED with an albumin of 1.0 has an even dimmer prognosis than someone coming in with an albumin of 2.5.  Either way, this study should urge us to be wary of under-prescribing empiric antimicrobials for septic older adults admitted with a lower serum albumin. 30680575

Time to blood culture positivity predicts both endocarditis and mortality in S. aureus bacteremia.  This makes sense to me:  a higher burden of infection equals more organisms in the blood drawn for culture, which equals fewer doubling times needed to generate a sufficient number of bacteria to be detected by the blood culture bottle system.  The authors prospectively studied all patients with S. aureus bacteremia across four university hospitals reporting culture time to positivity (TTP) over a two-year period. The outcomes of interest were diagnosis of definite endocarditis by Duke-Li criteria and 30-day mortality.

The study included 587 patients with S. aureus bacteremia (mean age 65; 72% male), of whom 7% were found to have definite IE and 21% died.  The median TTP of the initial positive blood culture was 13.7 hours.  Endocarditis was more common among patients with a TTP <10hr (12% vs 6%; p=0.002).  Of note, the authors describe endocarditis risk as having a U-shaped curve with TTP and also being more likely with a TTP >18 hours, but I redid the math with the numbers they presented in the abstract and that doesn’t seem to be the case. But maybe I’m missing something.  Anyway, 30-day mortality was also associated with shorter TTP (25% vs 16% in patients with a TTP <13.7hr vs >13.7hr; p=0.02).  Other factors associated with mortality included age, methicillin-resistance, stroke, pneumonia, and CRP.

Long story short – in this large prospective cohort of patients with S. aureus bacteremia, 60% of patients who died had a TTP <13.7hr, and 40% of patients with definite endocarditis had a TTP <10hr, suggesting that this readily available information from the micro lab had modest predictive value. 30036664

FDG-PET scans are valuable in the diagnosis of prosthetic valve, but not native valve, endocarditis.  This retrospective cohort study out of Brazil assessed the outcomes of 303 episodes of suspected left-sided endocarditis evaluated by FDG-PET (188 prosthetic valve / aortic graft, 115 native valve) over a three-year period.  The additional diagnostic workup for each patient included TTE and/or TEE, three sets of blood cultures, and cultures of explanted valves for the patients undergoing surgery; the final (i.e. gold standard) diagnosis was established by the hospital’s multidisciplinary endocarditis team, which did not use the FDG-PET data as part of their diagnosis.

Among the 188 patients with suspected prosthetic valve and/or aortic graft infection, FDG-PET had a sensitivity and specificity of 93% and 90%, respectivity; PPV and NPV were 89% and 94%.  Inclusion of FDG-PET would have reclassified 76% of the “probable” prosthetic valve/graft infections as “definite” infection.  Among the 115 patients with suspected native valve infection, FDG-PET had a sensitivity of only 22%; specificity was 100%, and accordingly the PPV and NPVs were 100% and 66%.

So, with a 93% sensitivity and 94% NPV, FDG-PET is very nearly an effective rule-out test for prosthetic valve and/or aortic graft infection (the NPV of a FDG-PET combined with a negative TTE or TEE is almost certainly >95%); alas, the same cannot be said for native valve infections. 30949690

Speaking of prosthetic device endovascular infections, the Mayo Clinic group just published this nice comparison of culture versus PCR testing of sonicate fluid for the diagnosis cardiovascular implantable electronic device (CIED) infection.  If “sonicate fluid” is a new term for you, the general idea is that you explant a device with suspected infection, put it in a machine called a sonicator that bombards it with sound waves to agitate and disrupt any biofilms or other particles attached to the device (think lithotripsy), and then culture or PCR any bacteria that have been shaken loose.  This approach has proven value in other device infections, like PJI.  Anyway, the authors reviewed a total 332 specimens that had sonicate fluid testing and assocted clinical data, stratified by whether CIED infection was ultimately deemed present.  Of these, 278 (86%) were ultimately deemed infected, and when comparing the results of sonicate culture versus PCR, sensitivity was higher with the latter (64% vs 58%; p=0.003), which identified a pathogen in 28/118 culture-negative cases, 18/28 of which represented staphylococci. So: explanted CEID sonication with PCR of the sonicate fluid seems like the diagnostic route of choice. 30944928