Predicting Risk of Drug-Resistant Pathogens in Patients with Pneumonia. (4th October 2017)

• Record Type: Journal Article
• Authors: Deshpande, Abhishek; Haessler, Sarah; Brizendine, Kyle; Richter, Sandra S; Lindenauer, Peter; Yu, Pei-Chun; Zilberberg, Marya D; Imrey, Peter; Higgins, Thomas; Rothberg, Michael
• Is Part Of: Open forum infectious diseases. Volume 4(2017)Supplement 1
• Page Start: S571
• Page End: S572
• Publication Date: 2017-10-04
• Subjects: Communicable diseases -- Periodicals; Medical microbiology -- Periodicals; Infection -- Periodicals; 616.9
• DOI: 10.1093/ofid/ofx163.1494
• Languages: English
• ISSNs: 2328-8957

Abstract

Abstract: Background: Antibiotic resistance is a major concern in hospital patients admitted with pneumonia. Clinicians need tools to identify patients at increased risk for MDRO infection who may require broad-spectrum antibiotic coverage. We derived a model for risk of pathogens resistant to community-acquired pneumonia (CAP) therapy, using a large sample of pneumonia patients with positive blood cultures. Methods: This cross-sectional study assessed adult patients admitted with pneumonia from 2010–2015 to 168 US hospitals that provided administrative and microbiologic data to Premier, Inc. We included all patients with positive blood cultures drawn by hospital day 1. We used stepwise multiple logistic regression to screen potential predictors of resistance to CAP therapy: sociodemographics (age, sex, race, marital status); admissions, antibiotics and drug resistance in the past year, health care–associated pneumonia (HCAP) risk factors, comorbidities, and severity of illness. For parsimony, we culled predictors based on prevalence, odds ratio, and clinical and statistical judgment. We developed our model in an 80% training set, reserving a 20% simple random sample reserved for model validation, and compared its predictive capability to those of the Drug Resistance in Pneumonia (DRIP) score, the HCAP definition, and a model using the HCAP definition's 4 components. Results: 5491 patients met inclusion criteria, and resistant pathogens were recovered in 1588 (28.9%). A 21-predictor stepwise model was further reduced to a 10-predictor model. Factors associated with antibiotic resistance in the reduced model were past year hospital admission, more so if within 2 months; admission from a skilled nursing facility; treatment with fidaxomicin or metronidazole within the past year; pressure ulcer; paralysis; comorbidity count; public payor; and male sex. Alcohol abuse and congestive heart failure predicted lower risk of resistance. In the validation set the c-statistic was 65.3% compared with 58.9% for the HCAP model, 58.0% for the DRIP score and 57.0% for the HCAP definition. Conclusion: Indicators readily available at admission predict antibiotic resistance with modest accuracy in pneumonia patients with positive blood cultures. The model slightly outperformed the HCAP model and the DRIP score. Disclosures: S. Haessler, AHRQ: Investigator, Research grant; S. S. Richter, bioMerieux: Investigator, Research support; BD Diagnostics: Investigator, Research support; Roche: Investigator, Research support; BioFire: Investigator, Research support; OpGen: Investigator, Research support; P. C. Yu, AHRQ: Investigator, Research grant; M. D. Zilberberg, Astellas: Consultant and Investigator, Research support; Merck: Consultant and Investigator, Research support; The Medicines Company: Consultant and Investigator, Research support; Shionogi: Consultant and Investigator, Research support; Pfizer: Consultant and Investigator, Research support; Theravance: Consultant and Investigator, Research support; J&J: Shareholder, Shareholder; M. Rothberg, AHRQ: Investigator, Research grant