For generations, a stethoscope has been one of the most essential tools for diagnosing pneumonia. But a new study suggests that what doctors hear through it may depend on who is listening.
In this cohort study, US investigators who are part of the Pediatric Emergency Care Applied Research Network analyzed data from children and teens diagnosed as having community-acquired pneumonia (CAP). The findings showed that clinicians often disagreed about their findings on physical exam, including decreased breath sounds, crackles (wet, bubble-likes sounds when air flow is blocked), and rhonchi (abnormal breathing that sounds like snoring or rattling)—symptoms frequently associated with pneumonia.
Pneumonia accounts for nearly 2 million pediatric outpatient visits and 375,000 emergency department visits each year in the United States. Current US guidelines recommend diagnosing CAP based on clinical findings rather than chest X-rays in children who are treated as outpatients.
Same patient, different findings
For the study, published in JAMA Network Open, researchers analyzed data from 252 participants, ages 3 months to 17 years, diagnosed with CAP at one of seven academic pediatric emergency departments across the country. Two clinicians independently examined each patient within 60 minutes of one another and recorded their findings.
None of the physical exam findings met the researchers’ predetermined threshold for acceptable interrater reliability, a measure of how much different clinicians agree when analyzing the same data or event. Wheezing (kappa value [κ], 0.50) and chest retractions (κ, 0.49) were the exam findings clinicians agreed on the most. Even so, agreement was modest.
Agreement was significantly lower for decreased breath sounds and crackles, which both had κ values under 0.25.
The findings raise questions about how heavily clinicians should rely on listening to the lungs when diagnosing pneumonia.
This variability is not a trivial concern.
“This variability is not a trivial concern,” write Susan Lipsett, MD, of Boston Children’s Hospital, and Mark Neuman, MD, MPH, of Harvard Medical School, in an accompanying commentary. When examination findings vary depending on the observer, “their utility as diagnostic anchors is diminished” and their ability to appropriately guide treatment is compromised.
“If physicians cannot consistently agree on the presence of auscultatory findings, treatment decisions may hinge more on examiner interpretation than underlying pathology,” Lipsett and Neuman write. “This variability may contribute to well-documented differences in antibiotic prescribing and chest radiograph use across institutions.”
Why lung sounds can be hard to interpret
Several factors could explain the disagreement. Accurately interpreting lung sounds can depend on background noise, patient cooperation, and clinician experience—factors that “become even more challenging in a busy emergency department,” write Lipsett and Neuman. Children also have more flexible chest walls, which may make it harder to determine exactly where the sounds originate. What’s more, clinicians may also use terms like “crackles” to refer to slightly different things.
The commentators argue that the results strengthen the case for objective risk-assessment tools that combine factors such as fever, oxygen saturation, demographics, and selected clinical findings rather than relying on individual lung sounds.
“By quantifying risk and reducing dependence on subjective auscultatory interpretation, clinical prediction tools may mitigate interobserver variability and promote more consistent decision-making regarding imaging and antibiotic therapy,
