2234. Expanding the WHO-Defined Patterns of Pneumonia on Pediatric Chest Radiograph to Better Guide Diagnosis and Appropriate Antibiotic Use
Authors * Denotes Presenting Author
  1. Denise Castro; Kingston Health Science Centre, Queen's University
  2. Michael Clay Kurtz *; Kingston Health Science Centre, Queen's University
  3. Wilma Hopman; Kingston Health Science Centre, Queen's University
  4. Siddharth Mishra; Kingston Health Science Centre, Queen's University
  5. Kirk Leifso; Kingston Health Science Centre, Queen's University
  6. Don Soboleski; Kingston Health Science Centre, Queen's University
The purpose of this study was to expand on previous WHO-defined radiographic descriptions of pneumonia to allow for the observation of bilateral symmetric versus asymmetric patterns of infiltrates to guide appropriate use of antibiotics.

Materials and Methods:
A retrospective study of 978 patients divided into age groups 3 months < 2y, 2-<4y, 4-<6y and 6y+ presenting to the emergency room of one of two hospitals who underwent chest x-ray (CXR) in their workup for query pneumonia. Two pediatric radiologists independently reviewed the CXRs blinded to clinical data. Each case was interpreted as one of the following patterns: normal (N); symmetric interstitial (SI); asymmetric interstitial (AI); asymmetric airspace (AAs); mixed (M) interstitial/airspace. Symmetry of infiltrates was determined by comparing the degree of overall aeration of each lung. Due to rarity within our age groups, a symmetric airspace pattern was not included. Further data included lung volumes (increased vs. decreased), presence of pleural fluid, lymphadenopathy or atelectasis. Extensive clinical data were obtained, including identified organism when available, correlated with CXR pattern.

Overall, 42% of CXRs were interpreted as normal, 19% as SI, 10% as AI, 18% as AA, and 11% as M. Increasing age was associated with an increased percentage of normal CXR interpretations. Increasing age was also associated with decreasing SI, AI, and M patterns and increasing AA pattern percentage (p < 0.001). Known respiratory syncytial virus (RSV) positive cases had normal interpretations in 20%, SI pattern in 46%, AI in 16%, and M pattern in 19% (p < 0.001). There were no known RSV cases demonstrating an AAs pattern in our study. Pleural effusions and lymphadenopathy were rare and significantly associated with an AAs pattern (p < 0.001).

Previous studies have associated AA CXR patterns with bacterial processes requiring antibiotic treatment and interstitial patterns with viral. Despite this observation, inappropriate antibiotic use is rampant in children presenting with pneumonia, contributing to an increase in drug resistance. A typical viral process should result in a more diffuse pneumonic process compared to a bacterial one. The distribution of pneumonia patterns in our study, including the group with known RSV, along with the change in CXR pattern percentages with age, mirrors the epidemiology literature regarding expected incidence of viral versus bacterial versus mixed infectious pneumonic processes. Our study expands on the WHO-defined CXR pneumonia patterns to include the observation of a symmetric interstitial nature to the infiltrate to support a viral process. We hope this observation will help support the physician’s decision not to prescribe antibiotics for the child with pneumonia. A significant percentage of viral pneumonias will have normal radiographs which in any case should also not warrant antibiotic use. The group of patients with an AI pattern understandably may receive antibiotic treatment based on CXR findings.