Related Experiment Videos
Center differences in a study of the efficacy of B-scan ultrasound
1Department of Radiology, New England Deaconess Hospital, Boston, Massachusetts.
Insights
This study found that variations between medical centers can significantly impact multi-center trial results, even with standardized protocols. Careful consideration of center differences is crucial for accurate carotid atherosclerosis research.
Area of Science:
- Medical Imaging
- Cardiovascular Research
- Clinical Trials
Background:
- Carotid atherosclerosis assessment is vital for stroke risk evaluation.
- Standardized imaging protocols are essential for multi-center trials.
- Evaluating variations in imaging interpretation is necessary for reliable results.
Purpose of the Study:
- To assess the efficacy of B-mode ultrasound for carotid atherosclerosis.
- To evaluate within-reader and between-reader variation in imaging studies.
- To determine the impact of multi-center differences on trial outcomes.
Main Methods:
- A four-year NIH-sponsored multi-center trial involving 234 vessels.
- Multiple readings of B-scan ultrasound and angiographic studies.
- Adherence to a common protocol monitored centrally and onsite.
Main Results:
- Within-reader correlation for minimal residual lumen: angiography (0.85-0.99), B-scan (0.62-0.93).
- Between-reader correlation for minimal residual lumen: angiography (0.50-0.98), B-scan (0.59-0.87).
- Significant variations observed across centers for multiple measurements.
Conclusions:
- Center differences can substantially influence multi-center trial results.
- Standardized protocols may not fully mitigate center-specific variations.
- Further research is needed to address inter-center variability in clinical trials.
Abstract:
As part of a four year NIH-sponsored multi-center trial to evaluate the efficacy of B-mode ultrasound imaging of carotid atherosclerosis, multiple readings of B-scan and angiographic studies of 234 vessels were obtained to permit intramodality assessment of within-reader and between-reader variation. All readers were obligated to adhere to a common protocol that was monitored both onsite and at a central coordinating facility. Among the five participating centers, the within-reader correlation coefficients for measurement of minimal residual lumen ranged from 0.85 to 0.99 for angiography and 0.62 to 0.93 for B-scan. Between-reader correlation coefficients for minimal residual lumen varied from 0.50 to 0.98 for angiography and 0.59 to 0.87 for B-scan. Analysis of center differences for other measurements demonstrated similar variations. This suggests that center differences may have a significant impact upon results of multi-center trials even when identical protocols are utilized.