When a patient presents at the hospital with a low-grade fever and acute cough, the physician on duty suspects pneumonia. Although the X-ray comes back negative, a computed tomography scan suggests an infiltrate. It's possible that the patient has another cough or viral-related illness, but the treating hospitalist diagnoses pneumonia and prescribes antibiotics.
This common scenario is leading to overdiagnosis of pneumonia and overuse of antibiotics, some experts say. Diagnosing pneumonia in the absence of some of the classic symptoms or a confirmatory X-ray contributes to drug resistance, increases health system costs, and unnecessarily exposes patients to drug side effects.
“There is a huge clinical overlap between community acquired pneumonia [CAP] and many other conditions with similar symptoms, such as viral infections that cause things like acute bronchitis, upper respiratory infection, or the common cold,” said Joshua P. Metlay, MD, PhD, FACP, chief of general medicine at Massachusetts General Hospital in Boston and an expert on pneumonia.
“There is strong evidence to support the use of antibacterial drugs for treatment of patients with pneumonia, but there's equally strong evidence that we shouldn't use those antibiotics to treat these non-bacterial causes of cough illness,” he added.
Lack of standardization in diagnosis is a fundamental problem, said Vinay Prasad, MD, an oncologist in the medical oncology branch of the National Cancer Institute in Bethesda, Md. While the hallmark symptoms of pneumonia—fever, cough, sputum production, dyspnea and pleurisy with confirmatory chest radiography—are well known, physicians have different thresholds for making a diagnosis and often rely on clinical judgment.
“In day-to-day practice, there is a range of definitions that people practically employ,” said Dr. Prasad. “Because it's a fluid definition based on clinical suspicion, with perhaps some radiographic confirmation, it's easier to apply the label in cases that may be borderline or may not be pneumonia, and that provides a reason to give antibiotics.”
Pressure to diagnose
It is not uncommon for physicians to diagnose pneumonia without an infiltrate on X-ray, said Dr. Prasad in a recent commentary in the Cleveland Clinic Journal of Medicine. He cited a multi-hospital study showing that 21% of patients admitted with pneumonia and 43% of those treated as outpatients had negative chest radiographs.
The problem is exacerbated by newer technology like CT scans, which can detect very small areas of cloudiness in the lungs that may or may not be pneumonia but are sometimes used as the basis for a positive diagnosis, he said.
The best practice is to wait for a confirmatory X-ray before initiating treatment, but several other factors typically influence a physician's decision, said Christopher J. Graber, MD, MPH, an infectious diseases specialist and director of antimicrobial stewardship at the VA Greater Los Angeles Healthcare System.
“A lot of it has to do with guidelines that say pneumonia must be treated in a set amount of time,” he said (typically within 4 hours or before the patient leaves the ED). “Physicians can feel pressured to give antibiotics before all the studies are back to confirm the diagnosis. If you're an ED physician, you have a greater incentive [to not miss a pneumonia case] than to withhold antibiotic therapy.”
Physicians may be more prone to diagnose pneumonia in patients considered at higher risk for complications, such as elderly patients who present with a high fever or high white blood cell count, said Christopher Tymchuk, MD, PhD, a hospitalist and infectious diseases specialist at the University of California Los Angeles Medical Center.
Also, a physician may decide to start a patient on antibiotics to be safe, with the intention of halting treatment if repeat imaging comes back negative. But then the latter step doesn't always happen.
“Sometimes there may not be clear findings on X-ray, but you start treatment just in case,” said Dr. Tymchuk. “Often, once they're on antibiotics, the feeling is that they're getting better because they're being treated, and so they get a full 5 to 7 days of antibiotics.”
Certain quality measures also inadvertently encourage physicians to favor a diagnosis of pneumonia over another illness with similar symptoms, said Dr. Metlay.
“HEDIS [Healthcare Effectiveness Data and Information Set] measures specifically monitor how often patients are getting antibiotics when they are diagnosed with acute bronchitis,” he said. “The answer should be rarely, if ever, but if that same patient is diagnosed with pneumonia, the diagnosis would not be included in that measure.”
Pneumonia may also be a more attractive diagnosis than an upper respiratory infection, for example, because it allows the physician to take concrete action to relieve the patient's symptoms, he said.
Establishing a threshold
Perfection of pneumonia diagnosis is close to impossible, but hospitalists should particularly strive to avoid antibiotic prescriptions for patients at very low risk of the disease and its complications.
“I'm less worried about overtreatment when the probability of pneumonia is 50% or more than when it is more like 5% or 1%,” Dr. Metlay said. “It should depend a lot on the clinical presentation and how frail or elderly the patient is.”
If a patient has all the signs and symptoms of pneumonia except an X-ray confirmation, “the probability of pneumonia is close to 50% and, especially for frail or elderly patients, it is reasonable to treat with antibiotics,” Dr. Metlay said.
Another dilemma in pneumonia treatment is whether to admit the patient. Established scoring systems, such as the pneumonia severity index (PSI) and CURB-65, are useful tools, but the physician still has to rely on clinical judgment, said Dr. Tymchuk. (CURB-65 = Confusion; blood Urea nitrogen >19 mg/dL; Respiratory rate ≥30; systolic Blood pressure [BP] <90 mm Hg or diastolic BP ≤60 mm Hg; age 65 or older.)
“A person might have an intermediate score on either the PSI or CURB-65 criteria but otherwise look good clinically and be able to be sent home on antibiotics with close follow up,” he said. Comorbidities are an important factor in admission decisions, he added.
Conversely, some factors can override a patient's low risk score and justify admission, said Dr. Metlay. For example, a patient who is too sick to swallow oral antibiotics, someone who cannot access medications, or a frail, elderly patient may be considered for admission despite a low risk score, he said.
“In the hospital, ventilator-associated pneumonia [VAP] can be a source of pneumonia overdiagnosis,” said Dr. Prasad. “It's especially important to use antibiotics appropriately for VAP, as these antibiotics are typically much broader in their coverage than those prescribed for CAP.”
“VAP is often the first thing that pops into your head when you see fever and a high white blood cell count in a patient on a ventilator,” said Dr. Graber. “But in reality there may be other things going on, like \1related colitis, a drug reaction, or an underlying condition that's evolving.”
If VAP is wrongly diagnosed, Dr. Graber added, clinicians may prescribe unnecessarily broad therapy to patients who “either don't have pneumonia or may not be at risk for some of the organisms covered by those antibiotics.”
Reversing the trend
Educating clinicians is important for alleviating the problem of overdiagnosis, experts said.
“You can do some social marketing by pointing out the downsides of overprescribing, so that the accepted thing to do in your hospital is to only prescribe antibiotics when appropriate,” said Dr. Graber. “You need to educate providers as to what constitutes a true diagnosis and engage key stakeholders in developing guidelines for treatment.”
At the VA Greater Los Angeles Healthcare System, clinicians must reassess therapy after 3 days whenever they prescribe 2 of the most commonly used broad-spectrum antibiotics, said Dr. Graber. Dr. Graber is also participating in a research project to develop dashboards so hospitalists can compare their own prescribing patterns with their peers'.
Ideally, physicians would be able to follow a set of predictive factors for lung infection to decide which patients would benefit from treatment and who can be safely followed, said Dr. Prasad. To that end, randomized clinical trials are needed to compare antibiotics with best supportive care in appropriate patients, he said.
The development of rapid diagnostic tests may offer the best potential for making diagnosis more precise and avoiding unnecessary use of antibiotics, said Dr. Metlay. For example, researchers are working on a point-of-care blood test to measure the level of procalcitonin—a potential biomarker for pneumonia—that could help clinicians differentiate viral from bacterial causes of infection.
“We're treating people empirically based on signs and symptoms, not based on the results of laboratory tests,” Dr. Metlay said. “That's the pragmatic approach until we get new rapid diagnostic tests.”