A Clinical Prediction Tool for Extended-Spectrum Cephalosporin Resistance in Community-Onset Enterobacterales Urinary Tract Infection

Predicting antibiotic susceptibility in urinary tract infection with artificial intelligence-model performance in a multi-centre cohort

Objective

To develop an artificial intelligence model to predict an antimicrobial susceptibility pattern in patients with urinary tract infection (UTI).

Materials and methods

26 087 adult patients with culture-proven UTI during 2015-2020 from a university teaching hospital and three community hospitals in Hong Kong were included. Cases with asymptomatic bacteriuria (absence of diagnosis code of UTI, or absence of leucocytes in urine microscopy) were excluded. Patients from 2015 to 2019 were included in the training set, while patients from the year 2020 were included as the test set.Three first-line antibiotics were chosen for prediction of susceptibility in the bacterial isolates causing UTI: namely nitrofurantoin, ciprofloxacin and amoxicillin-clavulanate. Baseline epidemiological factors, previous antimicrobial consumption, medical history and previous culture results were included as features. Logistic regression and random forest were applied to the dataset. Models were evaluated by F1-score and area under the curve-receiver operating characteristic (AUC-ROC).

Results

Random forest was the best algorithm in predicting susceptibility of the three antibiotics (nitrofurantoin, amoxicillin-clavulanate and ciprofloxacin). The AUC-ROC values were 0.941, 0.939 and 0.937, respectively. The F1 scores were 0.938, 0.928 and 0.906 respectively.

Conclusions

Random forest model may aid judicious empirical antibiotics use in UTI. Given the reasonable performance and accuracy, these accurate models may aid clinicians in choosing between different first-line antibiotics for UTI.

Fever lasting 48 hours as a predictive factor of ESBL-producing bacteria in non-critically ill patients with urinary tract infection

Urinary tract infection (UTI) is the most prevalent urological condition worldwide. Choosing appropriate antibiotics for patients who have fever before receiving a culture result is challenging. This retrospective study enrolled patients 394 patients hospitalized at Gangneung Asan Hospital for UTI from May 2017 to April 2021. Fever at 48 h of hospitalization was the analysis point, as this is when the response to antibiotic therapy manifest, although the results of antibiogram are not available. Multivariate analysis was performed to assess the correlation between ESBL producing bacteria (EPB) and fever at 48 h. Overall, 36.3% of patients had EPB and 27.9% had fever at 48 h. In multivariate analysis, a significant positive association was found between EPB and fever (odds ratio 1.17, 95% CI 1.05-1.30, P = 0.004) Female had negative association with multivariate model (OR 0.83, 95% CI 0.73-0.94, P = 0.004). Diabetes did not demonstrate a significant association with EPB. (OR 1.10, 95% CI 0.99-1.22, P = 0.072). Fever at 48 h is associated with EPB and could be considered a predictive factor for EPB infection in patients with UTI. Antibiotic escalation may be considered in patients with fever at 48 h.

Repurposing antibiotic resistance surveillance data to support treatment of recurrent infections in a remote setting

In northern Australia, a region with limited access to healthcare and a substantial population living remotely, antibiotic resistance adds to the complexity of treating infections. Focussing on Escherichia coli urinary tract infections (UTIs) and Staphylococcus aureus skin & soft tissue infections (SSTIs) captured by a northern Australian antibiotic resistance surveillance system, we used logistic regression to investigate predictors of a subsequent resistant isolate during the same infection episode. We also investigated predictors of recurrent infection. Our analysis included 98,651 E. coli isolates and 121,755 S. aureus isolates from 70,851 patients between January 2007 and June 2020. Following an initially susceptible E. coli UTI, subsequent recovery of a cefazolin (8%) or ampicillin (13%) -resistant isolate during the same infection episode was more common than a ceftriaxone-resistant isolate (2%). For an initially susceptible S. aureus SSTI, subsequent recovery of a methicillin-resistant isolate (8%) was more common than a trimethoprim-sulfamethoxazole-resistant isolate (2%). For UTIs and SSTIs, prior infection with a resistant pathogen was a strong predictor of both recurrent infection and resistance in future infection episodes. This multi-centre study demonstrates an association between antibiotic resistance and an increased likelihood of recurrent infection. Particularly in remote areas, a patient's past antibiograms should guide current treatment choices since recurrent infection will most likely be at least as resistant as previous infection episodes. Using population-level surveillance data in this way can also help clinicians decide if they should switch antibiotics for patients with ongoing symptoms, while waiting for diagnostic results.

Clinical Risk Scores to Predict Nonsusceptibility to Trimethoprim-Sulfamethoxazole, Fluoroquinolone, Nitrofurantoin, and Third-Generation Cephalosporin Among Adult Outpatient Episodes of Complicated Urinary Tract Infection

Background

Clinical risk scores were developed to estimate the risk of adult outpatients having a complicated urinary tract infection (cUTI) that was nonsusceptible to trimethoprim-sulfamethoxazole (TMP-SMX), fluoroquinolone, nitrofurantoin, or third-generation cephalosporin (3-GC) based on variables available on clinical presentation.

Methods

A retrospective cohort study (1 December 2017-31 December 2020) was performed among adult members of Kaiser Permanente Southern California with an outpatient cUTI. Separate risk scores were developed for TMP-SMX, fluoroquinolone, nitrofurantoin, and 3-GC. The models were translated into risk scores to quantify the likelihood of nonsusceptibility based on the presence of final model covariates in a given cUTI outpatient.

Results

A total of 30 450 cUTIs (26 326 patients) met the study criteria. Rates of nonsusceptibility to TMP-SMX, fluoroquinolone, nitrofurantoin, and 3-GC were 37%, 20%, 27%, and 24%, respectively. Receipt of prior antibiotics was the most important predictor across all models. The risk of nonsusceptibility in the TMP-SMX model exceeded 20% in the absence of any risk factors, suggesting that empiric use of TMP-SMX may not be advisable. For fluoroquinolone, nitrofurantoin, and 3-GC, clinical risk scores of 10, 7, and 11 predicted a ≥20% estimated probability of nonsusceptibility in the models that included cumulative number of prior antibiotics at model entry. This finding suggests that caution should be used when considering these agents empirically in patients who have several risk factors present in a given model at presentation.

Conclusions

We developed high-performing parsimonious risk scores to facilitate empiric treatment selection for adult outpatients with cUTIs in the critical period between infection presentation and availability of susceptibility results.

A multicenter analysis of trends in resistance in urinary Enterobacterales isolates from ambulatory patients in the United States: 2011-2020

BACKGROUND

Urinary tract infections (UTIs), which are usually caused by bacteria in the Enterobacterales family, are a common reason for outpatient visits. Appropriate empiric therapy for UTIs requires an understanding of antibiotic resistance in the community. In this nationwide study, we examined trends in antibiotic resistance in urinary Enterobacterales isolates from ambulatory patients in the United States (US).

METHODS

We analyzed the antimicrobial susceptibility profiles (extended-spectrum beta-lactamase [ESBL]-producing phenotype and not susceptible [NS] to beta-lactams, trimethoprim/sulfamethoxazole [TMP/SMX], fluoroquinolones [FQ], or nitrofurantoin [NFT]) of 30-day non-duplicate Enterobacterales isolates from urine cultures tested at ambulatory centers in the BD Insights Research Database (2011-2020). The outcome of interest was the percentage of resistant isolates by pathogen and year. Multi-variable generalized estimating equation models were used to assess trends in resistance over time and by additional covariates.

RESULTS

A total of 338 US facilities provided data for > 2.2 million urinary Enterobacterales isolates during the 10-year study. Almost three-quarters (72.8%) of Enterobacterales isolates were Escherichia coli. Overall unadjusted resistance rates in Enterobacterales isolates were 57.5%, 23.1%, 20.6%, and 20.2% for beta-lactams, TMP/SMX, FQ, and NFT, respectively, and 6.9% had an ESBL-producing phenotype. Resistance to two or more antibiotic classes occurred in 16.4% of isolates and 5.5% were resistant to three or more classes. Among isolates with an ESBL-producing phenotype, 70.1%, 59.9%, and 33.5% were NS to FQ, TMP/SMX, and NFT, respectively. In multivariable models, ESBL-producing and NFT NS Enterobacterales isolates increased significantly (both P < 0.001), while other categories of resistance decreased. High rates (≥ 50%) of beta-lactam and NFT resistance were observed in Klebsiella isolates and in non-E. coli, non-Klebsiella Enterobacterales isolates.

CONCLUSIONS

Antimicrobial resistance was common in urinary Enterobacterales isolates. Isolates with an ESBL-producing phenotype increased by about 30% between 2011 and 2020, and significant increases were also observed in NFT NS Enterobacterales isolates. Resistance rates for all four antibiotic classes were higher than thresholds recommended for use as empiric therapy. Non-E. coli Enterobacterales isolates showed high levels of resistance to commonly used empiric antibiotics, including NFT. These data may help inform empiric therapy choices for outpatients with UTIs.

MALDI-TOF MS Based Bacterial Antibiotics Resistance Finger Print for Diabetic Pedopathy

Diabetes mellitus has become a major global health issue. Currently, the use of antibiotics remains the best foundational strategy in the control of diabetic foot infections. However, the lack of accurate identification of pathogens and the empirical use of antibiotics at early stages of infection represents a non-targeted treatment approach with a poor curative effect that may increase the of bacterial drug resistance. Therefore, the timely identification of drug resistant bacteria is the key to increasing the efficacy of treatments for diabetic foot infections. The traditional identification method is based on bacterial morphology, cell physiology, and biochemistry. Despite the simplicity and low costs associated with this method, it is time-consuming and has limited clinical value, which delays early diagnosis and treatment. In the recent years, MALDI-TOF MS has emerged as a promising new technology in the field of clinical microbial identification. In this study, we developed a strategy for the identification of drug resistance in the diagnosis of diabetic foot infections using a combination of macro-proteomics and MALDI MS analysis. The macro-proteomics result was utilized to determine the differential proteins in the resistance group and the corresponding peptide fragments were used as the finger print in a MALDI MS analysis. This strategy was successfully used in the research of drug resistance in patients with diabetic foot infections and achieved several biomarkers that could be used as a finger print for 4 different drugs, including ceftazidime, piperacillin, levofloxacin, and tetracycline. This method can quickly confirm the drug resistance of clinical diabetic foot infections, which can help aid in the early treatment of patients.

Third-Generation Cephalosporin Resistance and Associated Discordant Antibiotic Treatment in Emergency Department Febrile Urinary Tract Infections

STUDY OBJECTIVE

Third-generation cephalosporin-resistant (3GCR) Escherichia coli, Klebsiella pneumoniae, and Proteus mirabilis (EKP) are an increasingly common cause of community-onset urinary tract infections (UTIs) in the United States. The 3GCR antimicrobial resistance pattern in these Enterobacterales species is most commonly due to production of extended-spectrum β-lactamases. We sought to provide contemporary, emergency department (ED)-focused data on 3GCR-EKP UTI regional prevalence, presentation, antibiotic susceptibility, and empiric treatment patterns, and outcomes.

METHODS

We performed a retrospective cohort study of all adults admitted with a febrile UTI at 21 Kaiser Permanente Northern California EDs between January 2017 and June 2019. Inclusion criteria included fever; admitting diagnosis of UTI, pyelonephritis, or sepsis; and ED urine culture with greater than 100,000 colony-forming units/mL of an EKP species. 3GCR was defined as in vitro resistance to ceftriaxone, ceftazidime, or both. 3GCR-EKP cases were compared with non-3GCR-EKP controls for the following: demographics, comorbidities, presenting clinical features, urinary isolate antimicrobial susceptibility, treatment, and clinical outcomes. The primary outcome measure was the rate of discordant initial empiric antibiotic treatment (administered within 6 hours of ED arrival) when compared with antimicrobial susceptibility testing. Secondary outcomes included hospital length of stay and 90-day mortality, adjusted for comorbidities and severity of illness.

RESULTS

There were 4,107 patients (median age 73 years and 35% men) who met study inclusion criteria. Of these patients, 530 (12.9%) had a 3GCR-EKP urinary tract infection. The proportion of subjects possessing risk factors for a health care-associated or extended-spectrum β-lactamase infection was 92.8% of case patients and 86.1% of controls. When comparing 3GCR-EKP case and non-3GCR-EKP control isolates, ciprofloxacin susceptibility rates were 21% versus 88%, and piperacillin/tazobactam susceptibility rates were 89% versus 97%, respectively. Initial empiric antibiotic therapy was discordant with antimicrobial susceptibility testing results in 63% of case patients versus 7% of controls (odds ratio 21.0; 95% confidence interval 16.9 to 26.0). The hospital length of stay was longer for 3GCR-EKP case patients, with an adjusted mean difference of 29.7 hours (95% CI 19.0 to 40.4). Ninety-day mortality was 12% in case patients versus 8% in controls (adjusted odds ratio 1.56; 95% confidence interval 1.07 to 2.28).

CONCLUSION

In this large, 2017 to 2019 Northern California ED study, nearly 13% of febrile EKP UTIs requiring hospitalization were caused by 3GCR-EKP, and in these cases, initial empiric therapy was often discordant with antimicrobial susceptibility testing. 3GCR-EKP infections were associated with a longer hospital length of stay and higher 90-day mortality. Similar data from other regions and for outpatient UTIs are needed.

Development and validation of a multivariable prediction model for infection-related complications in patients with common infections in UK primary care and the extent of risk-based prescribing of antibiotics

BACKGROUND

Antimicrobial resistance is driven by the overuse of antibiotics. This study aimed to develop and validate clinical prediction models for the risk of infection-related hospital admission with upper respiratory infection (URTI), lower respiratory infection (LRTI) and urinary tract infection (UTI). These models were used to investigate whether there is an association between the risk of an infection-related complication and the probability of receiving an antibiotic prescription.

METHODS

The study used electronic health record data from general practices contributing to the Clinical Practice Research Datalink (CPRD GOLD) and Welsh Secure Anonymised Information Linkage (SAIL), both linked to hospital records. Patients who visited their general practitioner with an incidental URTI, LRTI or UTI were included and followed for 30 days for hospitalisation due to infection-related complications. Predictors included age, gender, clinical and medication risk factors, ethnicity and socioeconomic status. Cox proportional hazards regression models were used with predicted risks independently validated in SAIL.

RESULTS

The derivation and validation cohorts included 8.1 and 2.7 million patients in CPRD and SAIL, respectively. A total of 7125 (0.09%) hospital admissions occurred in CPRD and 7685 (0.28%) in SAIL. Important predictors included age and measures of comorbidity. Initial attempts at validating in SAIL (i.e. transporting the models with no adjustment) indicated the need to recalibrate the models for age and underlying incidence of infections; internal bootstrap validation of these updated models yielded C-statistics of 0.63 (LRTI), 0.69 (URTI) and 0.73 (UTI) indicating good calibration. For all three infection types, the rate of antibiotic prescribing was not associated with patients' risk of infection-related hospital admissions.

CONCLUSION

The risk for infection-related hospital admissions varied substantially between patients, but prescribing of antibiotics in primary care was not associated with risk of hospitalisation due to infection-related complications. Our findings highlight the potential role of clinical prediction models to help inform decisions of prescribing of antibiotics in primary care.