Study demonstrates the scale of inappropriate antibiotic prescribing in the US
CHICAGO — A new Northwestern Medicine study found only 13 percent of outpatient antibiotic prescriptions were appropriate, with 36 percent considered potentially appropriate.
The study, conducted with the University of Michigan and Harvard University’s Brigham and Women’s Hospital, provides the most extensive assessment of outpatient antibiotic prescriptions to date and demonstrates the scale of inappropriate antibiotic prescribing in the U.S.
Overuse of antibiotics contributes to the development of antibiotic resistance — a major public health concern — increases health-care costs and exposes patients to unnecessary side effects.
The study will be published today, Jan. 16, in the journal The British Journal of Medicine (BMJ).
It also found that 23 percent of outpatient antibiotic prescriptions were inappropriate and 28 percent were not associated with any diagnosis code at all — suggesting that the rate of inappropriate prescriptions may in fact be even higher.
The study used a novel classification scheme and is uniquely comprehensive. For the first time, the scientists evaluated all 91,738 diagnosis codes inICD-10 (the system used in the U.S. to code diagnoses) and categorized each for antibiotic appropriateness.They also examined all outpatient antibiotic prescriptions among a cohort of 19.2 million patients, irrespective of the reason or site of care.
“Most prior studies have looked at antibiotic prescribing for a particular condition or in a particular location — for example, antibiotic prescribing for acute bronchitis in the emergency department,” said co-author Dr. Jeffrey Linder, chief of general internal medicine and geriatrics in the department of medicine and the Michael A. Gertz Professor of Medicine at Northwestern University Feinberg School of Medicine. “This allowed us to take a broader look at the appropriateness of antibiotic prescribing than has been done before.”
Despite initiatives to curb the problem, a significant proportion of prescribed antibiotics are unnecessary. However, prior studies are limited in scope and largely out of date; in particular, most relied on diagnosis codes in ICD-9, although the system was replaced with ICD-10 in 2015.
In the current study, the scientists developed a new, comprehensive ICD-10-based classification scheme that determined whether each of the more than 90,000 diagnosis codes “always,” “sometimes” or “never” justified treatment with antibiotics.
“No one had gone through all available codes before,” Linder said.
The team then used the new scheme to evaluate 15.5 million outpatient antibiotic prescriptions filled in 2016 by a large cohort of privately insured U.S. children and non-elderly adults. The scientists assigned each prescription fill to one of four categories: either “appropriate,” “potentially appropriate,” “inappropriate” or “not associated with a recent diagnosis code.”
They found that just 13 percent of prescriptions were appropriate, 36 percent were potentially appropriate and 23 percent were inappropriate. They also found that 28 percent were not associated with any diagnosis code at all — suggesting that the rate of inappropriate prescriptions may in fact be even higher.
“This means that our prior methods of looking at antibiotic prescribing based on location or specific diagnosis code is missing a huge proportion of antibiotics,” Linder said.
Beyond highlighting the widespread overuse of antibiotics in the U.S., the study could also help facilitate future research; the authors note that the new classification scheme could be applied to any dataset using ICD-10 codes, providing a valuable tool for scientists.
Dr. Kao-Ping Chua, assistant professor of pediatrics at the University of Michigan Medical School C.S. Mott Children’s Hospital, was the study’s corresponding author. Dr. Michael A. Fischer, associate professor of medicine in the division of pharmacoepidemiology and pharmacoeconomics at Harvard University’s Brigham and Women’s Hospital, was a co-author.
The BMJ study was supported by a grant from the Agency for Healthcare Research and Quality (R01HS024930) and a contract from the Agency for Healthcare Research and Quality (HHSP233201500020I).