Risk Factors for Adverse Outcomes with Outpatient Parenteral Antimicrobial Therapy
Wang, A. J.; Elnakieb, Y.; Bae, E.; Monogue, M.; Cutrell, J. B.; Lehmann, C. U.; Yen, C.; Medford, R. J.
Show abstract
ObjectivesOutpatient parenteral antimicrobial therapy (OPAT) use has increased significantly as it provides safe and reliable administration of long-term antimicrobials for severe infections. Benefits of OPAT include fewer antibiotic or line-related complications, increased patient satisfaction, shorter hospitalizations, and lower costs. Although OPAT programs carefully screen patients for eligibility and safety prior to enrollment, complications can occur. There is a paucity of studies identifying predictors of clinical outcomes in OPAT patients. Here, we seek to identify baseline predictors of OPAT outcomes utilizing machine learning methodologies. MethodsWe used electronic health record data from patients treated with OPAT between February 2019 and June 2022 at a large academic tertiary care hospital in Dallas, Texas. Three primary outcomes were examined: 1) clinical improvement at 30 days without evidence of reinfection; 2) patient actively being followed at 30 days; and 3) occurrence of any adverse event while on OPAT. Potential predictors were determined a priori, including demographic and clinical characteristics, OPAT setting, intravenous line type, and antimicrobials administered. Three classifiers were used to predict each outcome: logistic regression, random forest, and extreme gradient boosting (XGBoost). Model performance was measured using AUC, F1, and accuracy scores. ResultsWe included 664 unique patients in the study, of whom 57% were male. At 30 days, clinical improvement was present in 78% of patients. Two-thirds of patients (67%) were actively followed at 30 days, and 30% experienced an adverse event while on OPAT. The XGBoost model performed best for predicting treatment success (average AUC = 0.873), with significant predictors including ID consultation and the use of vancomycin. The logistic regression model was best for predicting adverse outcomes (average AUC = 0.710). Risk factors for adverse outcomes included management in the home setting and the use of vancomycin, daptomycin, or piperacillin-tazobactam. ConclusionOutcomes of patients undergoing OPAT can be predicted with the use of easily-obtainable clinical and demographic factors. Patients requiring certain antimicrobial therapies, such as vancomycin or daptomycin, may derive less benefit from early hospital discharge and OPAT.
Matching journals
The top 2 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Validation of a Derived International Patient Severity Algorithm to Support COVID-19 Analytics from Electronic Health Record Data 93%
- Clinical Utility of Automatable Prediction Models for Improving Palliative and End-Of-Life Care Outcomes: Towards Routine Decision Analysis Before Implementation 92%
- Real-Time Electronic Health Record Mortality Prediction During the COVID-19 Pandemic: A Prospective Cohort Study 92%
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Distinguishing Admissions Specifically for COVID-19 from Incidental SARS-CoV-2 Admissions: A National Retrospective EHR Study 94%
- Development of an interactive open source software application (RadaR) for infection management / antimicrobial stewardship 92%
- COHD-COVID: Columbia Open Health Data for COVID-19 Research 91%
Similar papers in this journal
- Development and Validation of ‘Patient Optimizer’ (POP) Algorithms for Predicting Surgical Risk with Machine Learning 92%
- Implicit bias in Critical Care Data: Factors affecting sampling frequencies and missingness patterns of clinical and biological variables in ICU Patients 91%
- Impact of CancelRx on Discontinuation of Controlled Substance Prescriptions 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.