Clinical guidelines for childhood pneumonia rest on a cornerstone assumption — that two trained clinicians examining the same child will reach broadly similar conclusions. A multi-site prospective cohort study now puts that assumption under serious scrutiny, with implications for millions of pediatric ED visits every year where imaging is intentionally withheld to reduce radiation exposure.
The PedCAPS study enrolled 252 children aged 3 months to 17 years at seven academic pediatric emergency departments across the United States between August 2023 and May 2025. Each child received two independent physical examinations within a 60-minute window, allowing direct comparison of clinician findings. Interrater reliability (IRR) was quantified using Fleiss kappa, with an acceptable threshold set at a lower 95% confidence bound of κ ≥ 0.4. The cohort was carefully defined — all participants had confirmed lower respiratory tract infection signs, recent fever, and radiographic pneumonia where imaging was performed — while children with complicating conditions such as immunodeficiency, cardiac disease, or chronic pulmonary illness (excluding asthma) were excluded to isolate straightforward community-acquired pneumonia presentations.
The findings arrive at a genuinely inconvenient moment for pediatric guidelines. Current recommendations from major bodies like the IDSA and PIDS explicitly encourage diagnosis through physical examination alone in outpatient-eligible children, in part to minimize chest X-ray use. If the foundational exam findings that drive that diagnosis carry poor reproducibility between examiners, the diagnostic framework itself becomes structurally fragile. Prior adult studies have documented similar variability in auscultation findings such as crackles and wheezing, but pediatric data from prospective, multi-site designs have been sparse. This study's seven-site design strengthens external validity, though the academic ED setting may not reflect community practice reliability, where time pressure and training variation are typically greater. Whether improved reliability could be achieved through standardized examination protocols or point-of-care ultrasound as a complement remains an open and pressing research question.