Related Experiment Videos
Interobserver variability in the interpretation of chest roentgenograms of patients with possible pneumonia
1Department of Medicine, Dalhousie University, Halifax, Nova Scotia.
Background:
Previous studies have found considerable interobserver variability in the roentgenographic diagnosis of pneumonia. In this study we determined the impact of experience on interobserver variability in the interpretation of chest roentgenograms of patients with pneumonia and defined the characteristics of chest roentgenograms interpreted as pneumonia by all groups.
Methods:
The chest roentgenograms of 15 consecutive patients with a clinical and roentgenographic diagnosis of pneumonia were read by a convenience sample of 10 first-year and 20 fourth-year medical students, 21 medical residents, and 21 staff physicians from the department of medicine. Three board-certified radiologists served as the gold standard. The majority rule decision and logistic regression analysis were used to analyze agreement among the various groups of observers.
Results:
Agreement with the majority decisions of the gold-standard radiologists on the presence or absence of pneumonia for all observations of the various groups were as follows: original radiologist, 87%; first-year medical students, 59%; fourth-year medical students, 54%; medical residents, 66%; and attending staff, 72%. Chest roentgenograms with dense lobar or segmental opacities were generally interpreted as pneumonia by all observers. Patchy opacities caused major disagreements and were usually read as not being due to pneumonia. Bronchopneumonia was not diagnosed by any of the observers. Air bronchograms, atelectasis, and chronic obstructive lung disease were usually not recognized by the nonradiologists.
Conclusions:
There is considerable interobserver variability in the roentgenographic diagnosis of pneumonia. This variability does not improve with increasing experience. Dense lobar or segmental opacities are uniformly recognized as pneumonia while patchy opacities result in a variety of interpretations.
Insights
Interobserver variability in diagnosing pneumonia from chest X-rays is significant and does not improve with experience. Dense opacities are consistently identified, but patchy opacities lead to varied interpretations.
Area of Science:
- Radiology
- Medical Diagnostics
- Pulmonary Medicine
Background:
- Interobserver variability in pneumonia diagnosis via chest roentgenography is a known issue.
- This study investigates how medical experience impacts this variability.
- It also defines roentgenographic features of pneumonia consistently identified by observers.
Purpose of the Study:
- To assess the effect of clinical experience on interobserver variability in interpreting chest roentgenograms for pneumonia.
- To identify specific roentgenographic characteristics of pneumonia that are consistently interpreted across different experience levels.
Main Methods:
- Chest roentgenograms from 15 pneumonia patients were reviewed by medical students, residents, and staff physicians.
- A panel of three radiologists served as the gold standard for diagnosis.
- Agreement was analyzed using majority rule and logistic regression.
Main Results:
- Observer agreement with the gold standard varied by experience level: original radiologist (87%), attending staff (72%), residents (66%), fourth-year students (54%), and first-year students (59%).
- Dense lobar or segmental opacities were consistently interpreted as pneumonia.
- Patchy opacities led to significant disagreement, often misinterpreted as not pneumonia.
Conclusions:
- Considerable interobserver variability exists in the roentgenographic diagnosis of pneumonia.
- Increasing clinical experience does not reduce this diagnostic variability.
- While dense opacities are uniformly recognized, patchy opacities present diagnostic challenges.