February 2019: ID diagnostics

When foot osteomyelitis recurs, the microbiology has usually shifted.  First, a disclaimer: this paper is from my home institution, and three out of the four authors are my mentors.

Foot osteomyelitis treatment fails about a quarter of the time.  The researchers, noting that the flora of these infections are diverse, asked whether treatment failures might be due to pathogens that either were missed on initial cultures or migrated into the infected area after treatment initiation.  To answer this question, the researchers retrospectively identified cases of foot osteomyelitis treatment failure (defined as unplanned resection of additional bone or above-ankle amputation after the initial surgery) occurring at a single VA hospital over a five-year period.  For reference, the standard approach to foot osteomyelitis at this hospital consists of surgical resection of grossly affected bone with intraoperative bone cultures, followed by 2-12 weeks of antibiotic therapy, preferentially oral, based on the organisms identified in the operative cultures.

Paired microbiologic data from bone samples obtained during the first and subsequent operations were available in 35/55 cases of foot osteomyelitis treatment failure.  The initial bone cultures identified 70 bacterial isolates and one fungal isolate; the subsequent bone cultures identified 77 bacterial and three fungal isolates.  Overall concordance between the first and subsequent cultures was poor. Enterococcus spp. were one of the most common organisms isolated and frequently persisted in the reoperation culture; Streptococci was also common in the initial culture but rarely seen in the subsequent culture.  On the other hand, S. aureus, P. aeruginosa, other Gram-negative organisms, and anaerobes were all frequently isolated for the first time in the reoperation culture.  These data suggest that in patients who fail an initial course of therapy for foot osteomyelitis, the subsequent treatment regimen should have activity against S. aureus, P. aeruginosa, and anaerobes. 30719591


Could we identify people at risk for prosthetic joint infection (PJI) with enough accuracy to justify antibiotic pretreatment?  I know this sounds like the infectious disease version of precrime, but hear me out. The authors sought to develop a clinical risk scoring system for postsurgical PJI occurring within the first three months after surgery. They conducted a multicenter prospective cohort study of patients undergoing hip and knee arthroplasties and randomly assigned them to derivation and validation cohorts. They created three models that included data collected prior to the arthroplasty, during the perioperative period, and at discharge, and determined the models’ diagnostic utility using the validation cohort.

The derivation and validation cohorts included 2324 and 1245 patients, respectively, among whom 43 (1.9%) and 22 (1.8%) had a PJI.  Median times to PJI diagnosis were 31 and 38 days in the derivation and validation cohorts.  The three models are presented in the text, with clinical scoring systems derived from multivariate analysis and their diagnostic parameters (sensitivity, specificity, NPV, and PPV) listed at various score thresholds. What I found most helpful here was that for each score cutoff the authors state what portion of the total population had that score.  So for example, the most powerful predictive model was the at-discharge model, which included in its risk score whether the arthroplasty was a total hip, whether it was a revision, whether the patient was obese, whether the patient’s National Nosocomial Infection Surveillance (NNIS) score was >2, whether the patient had a superficial wound infection, and whether the surgical wound was “high risk” – that is, whether it had bleeding, drainage, dehiscence or necrosis.  A score cutoff of >16 out of a possible 31 points, which included 1% of patients in both the derivation and validation cohorts, was 20-25% sensitive and 99% specific for subsequent PJI (PPV 38-41%, NPV 98-99%).

When you are able to winnow the population of potential infections to this degree – i.e. select out the 1% of folks undergoing arthroplasty at highest risk for PJI, such that 40% of the group you selected do in fact go one to have these complicated and devastating infections, then why not consider treating them preventatively? Say we give a few weeks of something that kills S.aureus and biofilm, like either minocycline or trimethoprim-sulfa plus rifampin, until the surgical wound has had a chance to close up and render the joint a closed sterile space once more.  If this antibiotic course was 100% effective in preventing PJI and you used the risk score thresholds above, you’d have a number needed to treat of about 3, and would actually be decreasing total antibiotic exposure in your patient population by getting rid of the extended courses of antibiotics patients are put through once they develop PJI.  This is all to say I’d really like to see and/or participate in a randomized controlled trial on the subject soon. 29715553

Procalcitonin can be used to safely withhold antibiotics in patients with acute COPD exacerbation.  Not terribly surprising – consistently, the best evidence for PCT as a predictor of infectious outcome and potential benefit of antibiotics has been in respiratory infections. But it’s always nice to have more data.   The authors retrospectively studied patients admitted to their hospital with acute COPD exacerbations and a PCT <0.25 ug/L, stratified by total duration of antibiotics received (<24 vs >24 hours).  Notable exclusionary criteria included presence of second infection requiring antibiotics, neutropenia or cell-mediated immunodeficiency, ICU admission, and radiographic evidence of pneumonia.  The outcome of interest was all-cause 30-day readmission.

A total 356 patients were included in the analyses, of whom 161 received less than 24 hours of antibiotics and 195 more than 24 hours of antibiotics.  The patients were in their sixties, two thirds were women, and one third used oxygen at home; overall the groups were similar.  In the <24hrs group, two-thirds of patients received no antibiotics at all; in the >24hrs group, the mean duration of therapy was 5.5 days.  There were no differences in all-cause 30-day readmission rates (16% vs 17%) or COPD-related 30-day readmission rates (11% vs 12%) between the <24hr and >24hr antibiotic groups, but the <24hr antibiotic group received a lower daily dose of corticosteroids (101 vs 117 predisone-mg equivalents; p = 0.02) and had shorter lengths of stay (2.8 vs 3.7 days; p=0.001).  Furthermore, when the authors compared the patients who had a hospital readmission (n=59) to those who did not (n=297), they found no difference in the durations of antibiotics prescribed (3.5 vs 3.1 days; p=0.4).

Take home point: in an immunocompetent patient with acute COPD exacerbation who has no radiographic evidence of pneumonia and can safely be admitted to the floor, antibiotics can safely be withheld (or stopped after the first dose) if the PCT is <0.25ug/L. 29982350


PCR of pleural fluid improves the diagnostic yield in pneumonias with parapneumonic effusions.  This study combined a 16s rRNA PCR assay to conventional culture techniques to detect pathogens from pleural fluid in patients with community-acquired pneumonia and parapneumonic effusions.  The authors prospectively studied adults admitted to one of four Swedish hospitals with CAP and pleural effusions over a three-year period.  As a control group, the PCR and cultures were also run on pleural fluid from 27 patients presenting to pulmonary clinic for thoracentesis; most had either malignant pleural effusions from various cancers or recurrent effusions due to heart failure.

One hundred ninety-seven patients were included in the study.  The mean age was 66 and men outnumbered women two to one; 91% had received a dose of antibiotic prior to their thoracentesis, and the mean duration of antibiotics prior to thoracentesis was 5.4 days.  Ninety-nine patients (50%) had a total 125 bacterial pathogens identified by either PCR, culture, or both.  The most common organisms were Viridans streptococci (30% - most of which were in the streptococcus anginosus group), Fusobacterium (14%), other anaerobes (14% - mostly Porphyromonas spp. and Prevotella spp.), and pneumococcus (11%).  PCR more often identified bacteria (in 62% vs 17% of the 152 pleural fluid samples submitted for both tests; p<0.0001) and more often identified multiple bacteria (12% vs 3%; p=0.003).  In the cases with empyema (described in the as “complicated parapneumonic effusion” – and defined by pleural fluid labs, observation of frank pus, or need for decortication), 83% of cases had one or more identified bacteria.  Of these, 30% were positive on culture, whereas 79% were detected on PCR.  No pathogens were detected by culture or PCR in the control group.

So, to sum up: half of patients with a parapneumonic effusion have bacteria in the pleural fluid, and while the yield of culture in such cases is low, adding PCR increases the yield by 2-4 fold.  Now of course, blind 16s rRNA is expensive, but a few organisms account for the majority of cases, so it seems like a multipathogen pleural fluid PCR panel would be a reasonable alternative. 30547334