Ingrid Hein

May 22, 2017

WASHINGTON — Nearly one in four patients treated with antibiotics for community-acquired pneumonia required additional antibiotic therapy, had to be hospitalized, or ended up in the emergency department, according to a study of more than 250,000 patients.

"We found it very surprising how frequently treatment fails," said investigator James McKinnell, MD, from the Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center in Torrance, California.

"Doctors need to be more aware of what they're tackling; community-acquired pneumonia is not to be trifled with," he told Medscape Medical News.

In their retrospective cohort analysis of outpatient community-acquired pneumonia, the investigators reviewed data in the MarketScan Commercial & Medicare Supplemental Databases from 2011 to 2015.

Dr McKinnell presented the results here at American Thoracic Society 2017 International Conference. Of the 251,947 cases identified, 55,741 patients (22.1%) needed further antibiotic treatment or ended up in the hospital.

Table 1. Follow-up Required After Treatment Failure for Community-Acquired Pneumonia

Follow-up n %
Antibiotic refill 11,493 20.6
Antibiotic switch 39,397 70.7
Emergency department visit 1835 3.3
Hospital admission 3015 5.4


Treatment for community-acquired pneumonia was more likely to fail if patients had at least one other medical condition.

Table 2. Comorbid Conditions Most Strongly Associated With Treatment Failure

Comorbid Condition Odds Ratio 95% Confidence Interval
Hemiplegia or paraplegia 1.33 1.17–1.51
Rheumatologic disease 1.28 1.21–1.35
Chronic pulmonary disease 1.25 1.21–1.29
Cancer 1.14 1.09–1.18
Diabetes 1.07 1.04–1.10
Asthma 1.05 1.01–1.10


Failure rates were similar, regardless of the class of antibiotic used.

Table 3. Treatment Failure by Antibiotic Class

Antibiotic Class Failure Rate, %
Beta-lactam 25.7
Macrolide 22.9
Tetracycline 22.5
Fluoroquinolone 20.8


Patients older than 65 years were nearly three times more likely to be hospitalized than younger patients, and nearly two times more likely after adjustment for risk.

Older patients are more vulnerable and should be treated more carefully, "potentially with more aggressive antibiotic therapy," Dr McKinnell explained. "This really shows that the current standard of care doesn't result in what we thought. We've got to get better at this."

There were significant regional variations in patient resistance to certain antibiotics; people on the East Coast did better than those on the West Coast, Dr McKinnell reported. "There might be less antibiotic resistance in that area."

Regional Differences in Drug Resistance

"Region ends up being a really important factor, he explained. "Different antibiotics have resistance in different parts of country, so that drives how your patients do."

"More fluoroquinolones are prescribed in some parts of country than others, so it follows that bacteria in those parts of the country will have different patterns of resistance." In this study, the data indicate that a broader antibiotic is needed on the West Coast.

"If you're not aware of outcomes in your area, you don't know what's driving outcomes," he said. "If we're going to use antibiotics, we've got to make sure we're using the right ones."

Physicians are usually not aware if the treatment they prescribed actually worked. "Often you don't get your patients coming back; they end up in emergency rooms, so primary care doesn't know that they did poorly," Dr McKinnell explained.

Big Data Will Guide Antibiotic Stewardship

This study — the largest of its kind — not only shows that more care is needed when prescribing antibiotics, but also that the way we make decisions needs to change.

"Clinical trials are great," he said, "but they don't give us information about what happens in the real world."

Big data has the potential to teach practitioners in the real world, outside of clinical trials, to make more accurate decisions for individual patients. "This is the start of antimicrobial stewardship — providing doctors tools to treat patients more effectively," he said. "It's a crying shame they don't already have that."

"We're told to be antibiotic stewards, but we don't have data to do that," Dr McKinnell said. "Why isn't the CDC doing this? Why aren't the payers doing it?"

In this study, his team was able to do this, with funding from the pharmaceutical industry. But really, Dr McKinnell pointed out, this work should be performed by the public sector.

"Big data can help drive better choices," he said. "This proves we need to move the realm of research to guide how we do medicine. This is an early foray into doing that, but we need more."

Current guidelines are helpful, but "we need to be more specific. Doctors need to track outcomes of their patients," Dr McKinnell said. Looking at the real performance of drugs in actual patients is the only way good decisions can be made. Big data "is starting to knock on that door."

"We need more robust data to actually make guidelines," said Shimshon Wiesel, DO, from Staten Island University Hospital, part of Northwell Health, in New York.

It is important to get national data on antibiotic failure, but "you can't make guidelines based on the national database," Dr Wisel told Medscape Medical News.

This is just the first step in the process, but "we need data that are more conclusive," he explained.

In the national database, you just get the diagnosis entered when the patient was seen. "It doesn't tell you how the patient presented, or what workup was done," he pointed out. The patient might have had bronchitis or a subtle presentation of lung disease; "these are known to fail on antibiotics," he said. "The biggest reason for treatment failure" is related to diagnostic criteria.

The study was funded by Truven. Dr McKinnell and Dr Wiesel have disclosed no relevant financial relationships.

American Thoracic Society (ATS) 2017 International Conference: Abstract 8450. Presented May 21, 2017.

Follow Medscape Pulmonary Medicine on Twitter @MedscapeLung and Ingrid Hein @ingridhein


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.