Decision-support models for empiric antibiotic selection in Gram-negative bloodstream infections.

OBJECTIVES: Early empiric antibiotic therapy in patients can improve clinical outcomes in Gram-negative bacteraemia. However, the widespread prevalence of antibiotic-resistant pathogens compromises our ability to provide adequate therapy while minimizing use of broad antibiotics. We sought to determine whether readily available electronic medical record data could be used to develop predictive models for decision support in Gram-negative bacteraemia. METHODS: We performed a multi-centre cohort study, in Canada and the USA, of hospitalized patients with Gram-negative bloodstream infection from April 2010 to March 2015. We analysed multivariable models for prediction of antibiotic susceptibility at two empiric windows: Gram-stain-guided and pathogen-guided treatment. Decision-support models for empiric antibiotic selection were developed based on three clinical decision thresholds of acceptable adequate coverage (80%, 90% and 95%). RESULTS: A total of 1832 patients with Gram-negative bacteraemia were evaluated. Multivariable models showed good discrimination across countries and at both Gram-stain-guided (12 models, areas under the curve (AUCs) 0.68-0.89, optimism-corrected AUCs 0.63-0.85) and pathogen-guided (12 models, AUCs 0.75-0.98, optimism-corrected AUCs 0.64-0.95) windows. Compared to antibiogram-guided therapy, decision-support models of antibiotic selection incorporating individual patient characteristics and prior culture results have the potential to increase use of narrower-spectrum antibiotics (in up to 78% of patients) while reducing inadequate therapy. CONCLUSIONS: Multivariable models using readily available epidemiologic factors can be used to predict antimicrobial susceptibility in infecting pathogens with reasonable discriminatory ability. Implementation of sequential predictive models for real-time individualized empiric antibiotic decision-making has the potential to both optimize adequate coverage for patients while minimizing overuse of broad-spectrum antibiotics, and therefore requires further prospective evaluation. SUMMARY: Readily available epidemiologic risk factors can be used to predict susceptibility of Gram-negative organisms among patients with bacteraemia, using automated decision-making models.

Utility of prior cultures in predicting antibiotic resistance of bloodstream infections due to Gram-negative pathogens: a multicentre observational cohort study.

OBJECTIVES: Appropriate empiric antibiotic therapy in patients with bloodstream infections due to Gram-negative pathogens can improve outcomes. We evaluated the utility of prior microbiologic results for guiding empiric treatment in Gram-negative bloodstream infections. METHODS: We conducted a multicentre observational cohort study in two large health systems in Canada and the United States, including 1832 hospitalized patients with Gram-negative bloodstream infection (community, hospital and intensive care unit acquired) from April 2010 to March 2015. RESULTS: Among 1832 patients with Gram-negative bloodstream infection, 28% (n = 504) of patients had a documented prior Gram-negative organism from a nonscreening culture within the previous 12 months. A most recent prior Gram-negative organism resistant to a given antibiotic was strongly predictive of the current organism's resistance to the same antibiotic. The overall specificity was 0.92 (95% confidence interval (CI) 0.91-0.93), and positive predictive value was 0.66 (95% CI 0.61-0.70) for predicting antibiotic resistance. Specificities and positive predictive values ranged from 0.77 to 0.98 and 0.43 to 0.78, respectively, across different antibiotics, organisms and patient subgroups. Increasing time between cultures was associated with a decrease in positive predictive value but not specificity. An heuristic based on a prior resistant Gram-negative pathogen could have been applied to one in four patients and in these patients would have changed therapy in one in five. CONCLUSIONS: In patients with a bloodstream infection with a Gram-negative organism, identification of a most recent prior Gram-negative organism resistant to a drug of interest (within the last 12 months) is highly specific for resistance and should preclude use of that antibiotic.

Impact of infection with extended-spectrum ß-lactamase-producing Escherichia coli or Klebsiella species on outcome and hospitalization costs.

BACKGROUND: Extended-spectrum ß-lactamase (ESBL)-producing bacteria are important sources of infection; however, Canadian data evaluating the impact of ESBL-associated infection are lacking. AIM: To determine whether patients infected with ESBL-producing Escherichia coli or Klebsiella species (ESBL-EcKs) exhibit differences in clinical outcome, microbiological outcome, mortality, and/or hospital resource use compared to patients infected with non-ESBL-producing strains. METHODS: A retrospective case-control study of 75 case patients with ESBL-EcKs matched to controls infected with non-ESBL-EcKs who were hospitalized from June 2010 to April 2013 was conducted. Patient-level cost data were provided by the institution's business office. Clinical data were collected using the electronic databases and paper charts. FINDINGS: Median infection-related hospitalization costs per patient were greater for cases than controls (C$10,507 vs C$7,882; median difference: C$3,416; P = 0.04). The primary driver of increased costs was prolonged infection-related hospital length of stay (8 vs 6 days; P = 0.02) with patient location (ward, ICU) and indirect care costs (including costs associated with infection prevention and control) as the leading cost categories. Cases were more likely to experience clinical failure (25% vs 11%; P = 0.03), with a higher all-cause mortality (17% vs 5%; P = 0.04). Less than half of case patients were prescribed appropriate empiric antimicrobial therapy, whereas controls received adequate initial treatment in nearly all circumstances (48% vs 96%; P < 0.01). CONCLUSION: Patients with infection caused by ESBL-EcKs are at increased risk for clinical failure and mortality, with additional cost to the Canadian healthcare system of C$3,416 per patient.

Duration of hospital admission and the need for empirical antipseudomonal therapy.

To reduce selective pressure for antimicrobial resistance, empirical use of antipseudomonal antibiotics is often reserved for patients with late-onset hospital-acquired infections. We examined the likelihood of isolating Pseudomonas aeruginosa as a function of time from hospital admission. We conducted a retrospective cohort study of all positive bacterial cultures in a tertiary-care hospital between March 2010 and November 2011. The primary outcome was the proportion of positive cultures yielding P. aeruginosa. Multivariable logistic regression was employed to assess the impact of time from admission on the likelihood of isolating P. aeruginosa, after adjusting for other important risk factors. A total of 7,668 positive cultures were obtained from 4,108 unique patients during the study interval, including 633 (8.3%) yielding P. aeruginosa. The probability of isolating P. aeruginosa increased linearly from 79/2,044 (3.9%) positive cultures obtained on admission to 153/664 (23%) in the 10th week of admission or beyond. The unadjusted odds ratio was 1.002/day (95% confidence interval [CI], 1.0016 to 1.0028; P < 0.0001); the adjusted odds ratio (aOR) was 1.0007/day (95% CI, 1.0001 to 1.0013; P = 0.02). Other important predictors of P. aeruginosa isolation included respiratory specimen type (aOR, 13.8; 95% CI, 9.1 to 21.1), recent hospital admission (aOR,1.8; 95% CI, 1.4 to 2.3), prior P. aeruginosa isolation during current admission (aOR, 4.9; 95% CI, 3.7 to 6.4), and prior antipseudomonal (aOR, 1.9; 95% CI, 1.4 to 2.5) or nonantipseudomonal (aOR, 1.8; 95% CI, 1.4 to 2.4) antibiotic exposure. It was determined that as time from admission increases, there is a linear increase in the likelihood of P. aeruginosa isolation. Any guidelines which distinguish early from late hospital-acquired infection must consider the implications of time point selection on the likelihood of inadequate P. aeruginosa empirical coverage.

Comparison of mortality between nosocomial and community-acquired febrile neutropenia patients treated initially with cefazolin plus tobramycin: retrospective chart review.

Cefazolin plus tobramycin have been determined to be effective for community-acquired FN, but have not been evaluated in the treatment of nosocomial FN. This study compared the incidence of mortality from 2002 to 2004 with 2008 to 2009 in patients with nosocomial FN treated with cefazolin plus tobramycin and compared characteristics of patients with nosocomially acquired FN to community acquired FN. A retrospective chart review of 45 nosocomial FN episodes from 2008 to 2009, and 54 episodes from 2002 to 2004 treated with cefazolin plus tobramycin was conducted. Data on the community acquired FN episodes was obtained from our previous research. Nosocomial FN mortality increased from 4% in 2002-2004 to 13% in 2008-2009 (p = 0.08). The nosocomial cohort was at higher risk of medical complications and mortality than the community-acquired cohort based on several variables (neutrophil nadir, duration of neutropenia and fever, hematological malignancy, MASCC and Talcott score; p < 0.05). As a result, the nosocomial cohort was treated with longer courses of antibiotic therapy (14 days vs 7 days; p < 0.0001) and were more likely to require broader spectrum antibiotics (64 out of 99 vs 34 out of 96; p < 0.0001). There was an observed increased risk of mortality from 2002 to 2004 compared with 2008 to 2009 in patients treated with cefazolin plus tobramycin for nosocomial FN, this was notable despite not attaining statistical significance. Therefore, this regimen is not appropriate for nosocomial FN.