Real-world evidence from EHR supports antimicrobial resistance fight

An automated system that extracts real-world data from the electronic health record is as efficient as manual data extraction in gathering real-world data to support research into antibiotic resistance, according to a new report on research supported by the US Food and Drug Administration (FDA).
 
The data extraction project, supported by FDA through the National Action Plan for Combating Antibiotic-Resistant Bacteria (CARB), began with investigators from the Johns Hopkins University School of Medicine building an automated system to extract patient-level data from electronic health records (EHRs) at five Johns Hopkins hospitals.
 
The researchers’ approach to pulling data from the Epic EHR “enhances access to the multicenter data needed to capture clinical practice and patient experience, streamlines analyses, and defends against [breaches] of patient privacy,” according to the “Regulatory Science in Action” report on the website of FDA’s Center for Drug Evaluation and Research (CDER). CDER lends support to the CARB research program.
 
According to the CARB national action plan, FDA has two aims in its role in the battle against antimicrobial resistance: to help develop new antibacterial drugs, and to advance the science of clinical trial design. The plan was launched in 2015 and last updated in October. Among the specific CARB research priorities for FDA is an aim to “evaluate potential innovations in clinical trial design for new antibacterial drugs such as enrollment strategies, data collection streamlining, drug development tools, clinical endpoints, and new statistical analytic approaches.”
 
The data extraction program developed by Johns Hopkins researchers compiled information about patient demographics, comorbidities, infection severity, lab results such as cultures and other microbiology data, and antibiotic treatments given. Researchers were also able to look for any information about sources and source control measures, and to track clinical outcomes.
 
Compared with manually collected data about these variables, “approximately 95% of the data elements needed to address study questions could be reliably captured using the automated extraction process,” according to FDA’s report.
 
Researchers were then able to use the automatically extracted data to look at optimal antibiotic treatments for two serious but common infections: bloodstream infections in adults caused by Pseudomonas aeruginosa, and pyelonephritis, or kidney infections, in children.
 
For P. aeruginosa infections, the investigators observed 249 adults who had positive blood cultures for the pathogen, finding no difference in death or 30-day recurrence of infection for a shorter compared with a longer antibiotic treatment course. Patients who received the shorter treatment course had an inpatient length of stay that was about 4 days shorter than those who receive longer courses of antibiotics. This study helped answer the question of whether an antibiotic course of under 10 days could achieve similar clinical outcomes to longer treatment courses for P. aeruginosa bloodstream infections.
 
In pediatric pyelonephritis, researchers wanted to know whether 7 days of antibiotic therapy would be as effective as a 2-week course. “[I]f so, the researchers proposed, the shorter treatment time would be greatly preferred for its potential to reduce the likelihood of antibiotic-associated adverse events such as Clostridioides difficile infection, end-organ toxicity, hypersensitivity reactions, and a furthering of antibiotic resistance.”
 
This second study looked at almost 800 children and found no difference in rates of recurrent urinary tract infections between those completing shorter and longer courses of antibiotics, with a trend toward fewer antibiotic-resistant recurrent infections in the short-treatment group.
 
“The approaches used in these studies for extracting patient data to compare treatment outcomes in real-world settings promise to assist FDA in future efforts to analyze large clinical databases,” according to this report. FDA hopes that a network of institutions can be built to participate in such efforts to optimize treatment of infectious disease and reduce antibiotic pressure and subsequent antibacterial resistance.
 
FDA