Search

Background Urine culture images collected using bacteriology automation are currently interpreted by technologists during routine standard-of-care workflows. Machine learning may be able to improve the harmonization of and assist with these interpretations. Methods A deep learning model, BacterioSight, was developed, trained, and tested on standard BD-Kiestra images of routine blood agar urine cultures from 2 different medical centers. Results BacterioSight displayed performance on par with standard-of-care–trained technologist interpretations. BacterioSight accuracy ranged from 97% when compared to standard-of-care (single technologist) and reached 100% when compared to a consensus reached by a group of technologists (gold standard in this study). Variability in image interpretation by trained technologists was identified and annotation “fuzziness” was quantified and found to correlate with reduced confidence in BacterioSight interpretation. Intra-testing (training and testing performed within the same institution) performed well giving Area Under the Curve (AUC) ≥0.98 for negative and positive plates, whereas, cross-testing on images (trained on one institution’s images and tested on images from another institution) showed decreased performance with AUC ≥0.90 for negative and positive plates. Conclusions Our study provides a roadmap on how BacterioSight or similar deep learning prototypes may be implemented to screen for microbial growth, flag difficult cases for multi-personnel review, or auto-verify a subset of cultures with high confidence. In addition, our results highlight image interpretation variability by trained technologist within an institution and globally across institutions. We propose a model in which deep learning can enhance patient care by identifying inherent sample annotation variability and improving personnel training.

Urinary stones are a global problem. While it has been established that microorganisms can be associated with urinary stone formation, data on best practices for microbial culture and clinical interpretation of microbial culture of urinary stone specimens are limited. Our objective was to investigate the microorganisms present in urinary stone cultures (usc) and companion urine cultures. We conducted a retrospective study at a tertiry hospital examining cultures submitted between October 2018 and October 2021. We collected data from electronic medical records of patients who had at least at least one usc performed. As part of routine clinical workup, all urinary stone specimens submitted for culture were ground using a disposable tissue grinder and inoculated on a tryptic soy agar with 5% sheep blood agar plate (BAP), a chocolate agar plate (CHOC), and a MacConkey agar plate (MAC), streaked for isolation and incubated in 5% CO2 at 35°C for up to three days. Urine cultures were performed using the Kiestra TLA system with 10 µL of inoculum onto each of a BAP and MAC, and plates were examined at 16 and 24 hours of incubation. For urine specimens submitted from a straight or “in and out” catheter, 10,000 CFU/mL was the threshold for workup. From clean-voided, midstream, ileostomy, nephrostomy, or indwelling/Foley catheter specimens, 100,000 CFU/mL was the threshold for workup. The final clinical significance of cultures is determined by evaluating the culture result in light of the patient’s clinical presentation. For usc and urine cultures, clinically insignificant growth was considered no growth or when no pathogens were recovered in culture. Urine cultures with more than 3 species over threshold were considered contaminated. A total of 1,049 usc were performed from 854 patients (55.3% female). The median patient age was 61 years (IQR 49-70). The most frequent microorganisms reported in usc cultures were Proteus mirabilis (N=98), Enterococcus faecalis (N=75), and Escherichia coli (N=73). No pathogen was identified from 57% of usc. 441 usc were performed in a context of urine culture (+/-2 days) in 336 (39.3%) patients. In addition, 138 patients (16.1%) had more than 1 usc, being the maximum as 11 usc performed in one patient. The most common urine/USC concordant result was no growth in 165 of 441 cultures (37.4%), followed by coagulase negative Staphylococcus. Cohen’s kappa obtained from usc and urine culture results was 0.45, which is not the minimum acceptable agreement. A pathogen was identified from usc in 43% of cases. Usc and urine cultures rarely resulted in concordant results, but the most common concordant result was no growth in culture. Additional studies are warranted to evaluate the impact of usc results on patient care, including antimicrobial use and correlation of microbes isolated with stone chemical composition.

Background Immunocompromised (IC) patients (pts) can have prolonged SARS-CoV-2 PCR positivity, even after resolution of COVID-19 symptoms. This study aimed to determine if viable virus could be detected in samples collected > 21 days after an initial positive (pos) SARS-CoV-2 PCR in IC pts. Methods We obtained 20 remnant SARS-CoV-2 PCR pos nasopharyngeal swabs from IC pts (bone marrow or solid organ transplant, high dose steroids, immunosuppressive medications) with a pos repeat PCR within the previous 30 days. The repeat specimens were cultured on Vero-hACE2-TMPRSS2 cells and incubated for 96 hours to assess viral viability. Viable RNA and infectious virus in the cultured cells were measured by qPCR and infectious plaque assays. RNA sequencing was performed on a HiSeq platform (Illumina). Samples also underwent SARS-CoV-2 antigen (Ag) testing (BD Veritor). Clinical data were extracted from the electronic health record by chart review. Results Pt characteristics are in Table 1. Viral cultures from the repeat specimen were negative (neg) for 18 pts and pos for 2 (Table 2). Pt 1 is a 60M treated with obinatuzumab 19 days prior to his first pos PCR test, with repeat specimen collected 21 days later (cycle threshold (Ct) not available). Pt 1 had a low viral titer (27 PFU/mL) & a D614G mutation on sequencing. Pt 2 is a 75M treated with rituximab 10 days prior to his first pos PCR test, with repeat specimen collected 23 days later (Ct 27.56/27.74). Pt 2 had a high viral titer (2e6 PFU/mL) and D614G, S98F, and S813I mutations. Demographics of Study Population (N=20) Characteristics of patients with a positive SARS-CoV-2 viral culture Conclusion 90% of specimens collected > 21 days after an initial pos SARS-CoV-2 PCR did not have viable virus detected on their repeat specimen. The 2 pts with pos viral cultures had active hematologic malignancies treated with an anti-CD20 mAb at the time of COVID-19 diagnosis. One pt had a high concentration of active, viable virus. No known variants of concern were noted in this cohort, collected in Q2 2020, though prolonged replication is a risk for variant development. Further data are needed about risk factors for persistent viable viral shedding & methods to prevent transmission of viable virus from IC hosts. Disclosures Victoria J. Fraser, MD, CDC Epicenters (Grant/Research Support)Cigna/Express Scripts (Other Financial or Material Support, Spouse is Chief Clinical Officer)Doris Duke Fund to Retain Clinical Scientists (Grant/Research Support, Research Grant or Support)Foundation for Barnes-Jewish Hospital (Grant/Research Support, Research Grant or Support)NIH (Grant/Research Support, Research Grant or Support) Victoria J. Fraser, MD, Centers for Disease Control and Prevention (Individual(s) Involved: Self): Grant/Research Support, Research Grant or Support; Cigna/Express Scripts (Individual(s) Involved: Spouse/Partner): Employee; Doris Duke Charitable Foundation (Individual(s) Involved: Self): Grant/Research Support, Research Grant or Support; National Institutes of Health (Individual(s) Involved: Self): Grant/Research Support, Research Grant or Support; The Foundation for Barnes-Jewish Hospital (Individual(s) Involved: Self): Grant/Research Support, Research Grant or Support Michael S. Diamond, MD, PhD, Carnival Corporation (Consultant)Emergent BioSolutions (Grant/Research Support)Fortress Biotech (Consultant)Immunome (Advisor or Review Panel member)Inbios (Consultant)Moderna (Grant/Research Support, Advisor or Review Panel member)Vir Biotechnology (Consultant, Grant/Research Support) Carey-Ann Burnham, PhD, BioFire (Grant/Research Support, Other Financial or Material Support)bioMerieux (Grant/Research Support)Cepheid (Consultant, Grant/Research Support)Luminex (Grant/Research Support)Roche (Other Financial or Material Support) Carey-Ann Burnham, PhD, BioFire (Individual(s) Involved: Self): Grant/Research Support; bioMerieux (Individual(s) Involved: Self): Grant/Research Support, Scientific Research Study Investigator, Speakers’ bureau; Cepheid (Individual(s) Involved: Self): Consultant, Grant/Research Support, Scientific Research Study Investigator; Luminex (Individual(s) Involved: Self): Scientific Research Study Investigator Hilary Babcock, MD, MPH, FIDSA, FSHEA, Nothing to disclose

BACKGROUND: Recurrent preschool wheezing is a heterogeneous disorder with significant morbidity, yet little is known about phenotypic determinants and their impact on clinical outcomes. OBJECTIVE: Latent class analysis (LCA) was used to identify latent classes of recurrent preschool wheeze and their association with future exacerbations and inhaled corticosteroid (ICS) treatment response. METHODS: Data from five clinical trials of 1,708 children age 12–71 months with recurrent wheezing were merged. LCA was performed on 10 demographic, exposure and sensitization variables to determine the optimal number of latent classes. The primary outcome was the annualized rate of wheezing exacerbations requiring systemic corticosteroids during the study intervention period; the secondary outcome was the time to first exacerbation. Exploratory analyses examined the effect of daily ICS treatment on exacerbation outcomes. RESULTS: Four latent classes of recurrent wheezing were identified; these were not distinguished by current symptoms or historical exacerbations but differed with regard to allergen sensitization and/or exposures. Annualized exacerbation rates (mean ± SEM/year) were 0.65 ± 0.06 for class 1 (“minimal sensitization”), 0.93 ± 0.10 for class 2 (“sensitization with indoor pet exposure”), 0.60 ± 0.07 for class 3 (“sensitization with tobacco smoke exposure”), and 0.81 ± 0.10 for class 4 (“multiple sensitization and eczema”) (p < 0.001). In a research setting of high adherence, daily ICS treatment improved exacerbation rates in classes 2 and 4 but not the other groups. CONCLUSION: Sensitization and exposure assessments are useful in the prediction of future exacerbation and may identify children most likely to respond favorably to daily ICS treatment.