A new analysis of Medicaid claims data has found that nearly 45% of antibiotics prescribed over a 10-year period were given out for no clear reason.
In a study published yesterday in Health Affairs, researchers with Brigham and Women's Hospital, Harvard Medical School, and Northwestern University Feinberg School of Medicine reported that, of the nearly 300 million antibiotic prescriptions filled by Medicaid recipients from 2004 through 2013, 17% were linked to clinician visits that didn't involve an infection-related diagnosis. Even more concerning was that more than one in four prescriptions had no record of any clinician visit.
The authors of the study say the findings are significant because they suggest that efforts to reduce inappropriate outpatient antibiotic prescribing in the United States may be missing a big part of the picture, since current strategies don't account for antibiotics prescribed outside of clinician visits. The country has one of the higher rates of antibiotic prescribing among high-income nations: more than 800 prescriptions per 1,000 Americans per year.
"We're not necessarily saying that all of these prescriptions are clinically inappropriate, but that's exactly the challenge that they raise," said lead author Michael Fischer, MD, a physician at Brigham and Women's and associate professor at Harvard Medical School. "It's so hard to judge if they are appropriate or not."
More than 82 million prescriptions with no clinician visit
For the study, Fischer and his colleagues looked for all outpatient antibiotic prescriptions filled by Medicaid patients from 2004 through 2013, then examined claims records to determine whether there was any kind of clinical encounter in the 7 days before the prescription was filled.
For all encounters identified within the 7-day window, the researchers looked for an infection-related diagnosis. If no encounters were identified within the 7 days before the prescription was filled, the researchers classified the prescription as non-visit-based.
Although the data set is old compared to other recent estimates of outpatient prescribing, Fischer said Medicaid recipients represent a truly national patient sample, one that provide a sense of how commonly antibiotics are prescribed without an infection-related diagnosis or a face-to-face encounter between doctor and patient.
Overall, the researchers identified 298 million antibiotic prescriptions filled by 53 million patients, with 62% of prescriptions written for children. While 72% of the prescriptions were associated with a clinician visit, 17% of those were not associated with an infection-related diagnosis, and 28%—more than 82 million prescriptions—had no claims for an encounter with a physician in the seven previous days. Only 4% of those antibiotic prescriptions were for dental procedures or were chronic prescriptions for conditions like acne—two prespecified subsets of prescriptions that are often prescribed without a clinician visit.
While there was a decrease in the proportion of non-visit-based antibiotic prescriptions over time, the percentage in 2013 was still substantial (22%). The proportion of antibiotic prescriptions with no infection-related diagnosis remained roughly the same over the study period.
With the non–infection-related prescriptions, Fischer suggested it's possible that the patients may have also had an infection, and that clinicians just didn't provide a diagnostic code for everything that came up during the visit.
For the non-visit-based prescriptions, the scenario in which these patients received antibiotics without a visit to the doctor is unclear, but Fischer and his colleagues assume that most of these prescriptions were given out over the phone—interactions that wouldn't be recorded in the claims data. And clinicians may have made notes about the prescription in the patients' charts, but that information wasn't available in the claims data.
Fischer said they did not anticipate that getting a prescription without a clinician visit would be as common as the results indicate.
"Again, we're not saying that the clinical care is inappropriate, but it's really hard to assess how well we're doing with the appropriateness of antibiotic use, and think about how we intervene on it, if we're not capturing the data accurately," he said.
J.D. Zipkin, MD, associate medical director of Northwell Health-GoHealth Urgent Care, a network of urgent care centers in New York, agreed.
"Unlike visits with complete medical notes, non-visit prescribing requires no associated diagnosis, making automated evaluation unattainable" said Zipkin, who was not involved in the study. "I suspect many of these prescriptions will be completely unsupported in a similar manner as in-visit inappropriate antibiotic prescribing."
More comprehensive measures of antibiotic use
For Fischer, the concern is what these findings mean for outpatient antibiotic stewardship efforts. Those efforts have focused on situations in which patients go to a provider with a cold or upper respiratory infection and receive an antibiotic—even though these conditions are mainly caused by viruses and don't require antibiotics. Previous studies have found that these conditions are the primary culprits in unnecessary antibiotic use.
But outpatient antibiotic stewardship strategies don't account for patients receiving antibiotics for non-infections, or getting them without even seeing a clinician. Not every clinic-based intervention can address these types of prescriptions, Fischer said.
"It suggests to me that we want to keep thinking about ways to more comprehensively reach out to both patients and clinicians to help change the way we look at antibiotic use," he said.
Better education about appropriate antibiotic use for both patients and providers will play a role, Fischer said. In addition, stewardship interventions will have to account for more "virtual" office visits, like telemedicine and patient portals.
"That would be an important area for people to be innovating," he suggested.
In the meantime, he added, researchers who are studying outpatient antibiotic use need to make sure they are capturing all the prescriptions being written, and getting the associated clinical data so they can determine whether the antibiotics were necessary.
"It's realizing that there's all this antibiotic prescribing that's outside of visits or without a visit clearly flagged for infection, and we need to think about how we're going to include those if we're going to measure total antibiotic use," Fischer said. "And for measuring appropriateness, we need the clinical data to do that."
Because the data set is old, Fischer said he hopes the findings will encourage other researchers to start looking at other patient populations, or more current Medicaid data, to see whether these trends have changed at all in recent years.
"Our hope is that lots of other people, whether they have slightly more current data, or more clinically detailed data, will keep digging into this phenomenon and get a more complete understanding of it," he said.
Feb 3 Health Affairs abstract