September 2019: ID Diagnostics

A “blood cultures last” approach to phlebotomy reduces the rate of blood culture contamination.  Over the past few years we’ve seen a few companies marketing blood culture collection devices that divert the first few microliters of the initial sample in order to reduce contamination with skin flora (e.g Kurin Lock, SteriPath, etc) with generally good results.  And I think that’s great!  But what if we could get similar results by changing practice rather than adding in (and paying for) a new layer of complexity to nurses’ and phlebotomists’ jobs?

The authors performed an open-label randomized controlled trial across two campuses at a single Israeli medical center between 2017 and 2018.  They included patients from whom the treatment team intended to collect blood cultures as well as other studies.  The intervention was simple: in the standard of care arm, the hospitals’ usual practices were maintained, and in the intervention arm, the nurse or phlebotomist was instructed to collect a green-top tube for the non-culture labs before collecting blood into the blood culture bottles.  Culture contaminants were defined as a single culture bottle growing one or more of coagulase-negative staphylococci, C. acnes, Micrococcus sp., viridans strep, aerobic diptheroids, or aerobic gram-positive rods.

In total, 756 non-duplicate blood cultures were collected from as many patients.  Half of patients had a discharge diagnosis of pneumonia, and 30% a diagnosis of UTI; Pitt bacteremia scores were similar between groups.  Most blood cultures were collected in the ED.  Rates of true bloodstream infection were no different between groups (6.5% with the intervention versus 6.3% with standard care), but the intervention resulted in a 66% reduction in the rate of culture contamination (1.7% vs 5.5%; OR 0.34 with 95% CI 0.14-0.94). 

I love new toys as much as the next person, but given this study I think it may make more sense for healthcare systems to start with this behavioral intervention, see how far down they can get the culture contamination rate, and then decide if adding these blood culture diversion devices is still cost-effective. 31539635

In hematopoietic stem cell transplant (HSCT) recipients, does viral PCR panel testing of the upper respiratory tract predict the results of the same testing in the lower respiratory tract?  This is an important question because HSCT recipients often develop lower respiratory tract (LRT) infections with viral pathogens, have severe outcomes from these infections, and accordingly are folks we occasionally decide to treat with ribavirin, IVIG, and even less proven therapies. 

The authors retrospectively examined HSCT recipients who presented to the Fred Hutchinson Cancer Center between 2009 and 2016 with respiratory symptoms and who received multiplex PCR testing for respiratory viruses on both upper respiratory tract (URT) and LRT samples, then examined the concordance between these tests.  They identified 235 patients who received both URT and LRT viral panel test within a 3-day period.  Of the 115 subjects who had one or more positive test results, testing was discordant in 42/115 cases (37%).  Of these, 40% had positive tests solely in the lower respiratory tract; this occurred most commonly with adenovirus, metapneumovirus, parainfluenza virus type III, and influenza A.  In contrast, URT and LRT testing was concordant for RSV in >90% of cases, and rhinovirus was frequently identified solely from the URT sample.  Nonetheless, a positive URT viral panel was strongly predictive of LRT viral panel positivity (OR 74 with 95% CI 27-204).

So, in a HSCT recipient it’s reasonable to trust that a positive viral PCR panel from an upper respiratory sample reflects the pathogens in the lower respiratory tract (particularly if it’s RSV; perhaps less so if it’s rhinovirus); however, if the upper tract test is negative and you still have a high index of suspicion, lower tract testing may be indicated. 31541573

Speaking of the multiplex PCR respiratory viral panel, a recent RTC involving 998 adults with respiratory symptoms, fever, chest pain, or “poor general condition” in Finland showed that viral testing resulting within 24 hours did not reduce antimicrobial consumption versus standard care (no rapidly resulting viral testing).  In total, 18% of the patients tested had a respiratory virus.  However, early detection of a viral pathogen did not affect the antibiotic durations prescribed (11.3 vs 10.4 days), mean length of stay (4.1 vs 4.2 days), total cost of hospital care (3539 vs 3339 euro), or rate of mortality (3% in each group) (p > 0.05 for all- and yes, I know p is a substandard way of communicating of statistical significance, but I am too lazy to copy out all of those confidence intervals).  This paper is yet one more to espouse a running theme: giving clinicians additional data without an active stewardship intervention does not actually change their antibiotic prescribing behaviors. 31574339

What is the cost, in terms of blood culture diagnostic yield, of giving antibiotics prior to blood cultures?  In September’s Annals of IM, Cheng et al report a prospective study of blood cultures from adults with severe sepsis presenting to one of seven North American EDs.  Here, severe sepsis was defined as SIRS plus either hypotension (SBP <90 mmHg) or an elevated serum lactate.  The participants had blood cultures collected prior to and up to 4 hours after the initiation of antimicrobial therapy.  The primary outcome of the study was the sensitivity of post-antibiotic blood cultures using the pre-antibiotic cultures as a reference standard.  Blood culture contaminants were defined as skin flora organisms present in a single set of paired blood culture, and each case of potential culture contamination was reviewed by two ID and clinical microbiology specialists.

A total 330 patients were enrolled in the study; 325 patients were included in the final analysis, and 264 were included in a subanalysis looking at cultures obtained 30-120 minutes after antibiotic administration.  The mean age was 66 years and men made up just over 60% of the cohort; a third of the patients had positive pre-antibiotic blood cultures, and the greatest predictors of pre-antibiotic blood culture positivity were abnormal temperature (60% vs 48%) and respiratory failure (20% vs 9%).  The mean time between antibiotic administration and collection of the second set of cultures was 70 minutes.  The absolute difference in rate of blood culture positivity between the pre- and post-antibiotic cultures increased with time after antibiotic administration, from 11% among the patients who had a second set collected within 30-60 minutes of receiving antibiotics to 21% among those cultured more than two hours after receiving antibiotics.  Put another way, the sensitivity of the post-antibiotic blood cultures was 53% overall, ranging from 64% when collected within 60 minutes of antibiotics to 42% when collected more than two hours later.  This data confirms the findings of prior retrospective studies: the diagnostic cost of delayed blood cultures is real and large. 31525774

Finally, a few short notes:

A review of 339 episodes of positive blood cultures with Corynebacterium found true in infection in 9% of cases.  That’s higher than I would have expected.  Anyway, the patients deemed to have true infection were mostly older men with multiple medical problems, and the species most commonly implicated in true infection were C.jeikeium and C.striatum.  Eight episodes of endocarditis were identified, and all involved prosthetic valves.  So, I guess in the patients for whom I worry about enterococcal bacteremia – to put it simply, old sick men with aftermarket parts – I should also be taking Corynebacterium bacteremia seriously. 31485919

Here’s a potential sea-change in tuberculosis therapy: whole genome sequencing (WGS) accurately predicts M.tuberculosis susceptibility to first-line drugs.  The authors examined nine genes and/or promoter region mutations related to resistance to rifampin isoniazid, pyrazinamide, and ethambutol across 1136 TB isolates submitted to The Netherlands’ national TB reference lab.  The found that WGS had a >99% negative predictive value for antitubercular resistance, and that implementation of WGS could reduce the need for traditional phenotypic testing by more than 90%.  This makes me wonder: how long before WGS can be performed directly on clinical samples (i.e. sputum) and reliably detect resistant TB isolates?  This seems like it would be game-changer for reducing the use of toxic second-line drugs in areas with high prevalences of MDR and XDR TB. 31119271

Circulating hepatitis C virus core antigen in plasma correlates well with HCV viral loads and may be an alternative to PCR-based diagnosis and monitoring of active HCV infection.  Compared to PCR, the antigen assay had a sensitivity of 97% and a specificity of 95% in a study of 303 samples from 124 patients either newly diagnosed with HCV or monitored during HCV treatment.  I guess the question is, what advantage does this new assay offer over the PCR – faster turnaround time?  Lower cost?  Point of care testing? 31580940

Of the two multiplex PCR assays for vaginitis on the market, the BD Max vaginal panel outperforms the BD Affirm VPIII microbial ID test.  When used to simultaneously test 200 symptomatic women, Max VP had a sensitivity and specificity of 96% and 96% for BV, versus Affirm’s 96% and 82%; for candida, Max VP had a sensitivity and specificity of 98% and 95% versus Affirm’s 69% and 100%; for trichomonas, both tests performed identically.  This study suggests Max VP should probably replace Affirm as the standard of care molecular diagnostic for vaginitis. 31502121