Author + information
- Anna Lisa Crowley, MD∗ (, )
- Eric Yow, MS,
- Dawn Rabineau, RDCS,
- Casey Norris, MS,
- Jennifer White, MS,
- Melissa A. Daubert, MD,
- Eric J. Velazquez, MD,
- Huiman Barnhart, PhD,
- Mitchell W. Krucoff, MD,
- Sunil V. Rao, MD and
- Pamela S. Douglas, MD
- ↵∗Department of Medicine, Duke University, Box 2805 DUM, Durham, North Carolina 27705
High-quality imaging data are critical when echocardiography is used for serial assessments of study and safety endpoints for regulatory decision-making. Although core laboratories (ECL) improve analytic consistency (1–3), site image quality is also required and was suboptimal in up to 50% for tissue Doppler imaging. (4) This variability in image acquisition represents a limitation to widespread use of echocardiography in this setting. In response, we tested the impact of a real-time quality program on image quality and measurability in the PRESERVATION-1 (Prevention of Remodeling of the Ventricle and Congestive Heart Failure After Acute Myocardial Infartion) trial (303 patients; 61 sites).(5)
To optimize the primary endpoint of diastolic left ventricular volume (LVV), we developed a quality program targeting structure, processes, and outcomes (Table 1). Site characteristics were compared with site baseline and quarterly measurability scores using median regression, with a primary quality endpoint of measurable paired LVV (baseline + 6 months).
Of 213 sonographers certified for 2D echocardiograms (2DE) by training and test echo submission, 102 (47.9%) were certified to perform 3DE on the first attempt. Thus, without additional quality initiatives, the trial primary endpoint of change in LVV would have been assessable by 3DE in no more than 47.9%. Instead, it was obtained in all 279 subjects with attempted baseline + 6-month echocardiograms (74% by 3D, 26% 2D; when both were available, 3DE and 2DE LVVs correlated closely: Pearson correlation coefficient 0.9998). Our quality program increased measurable 3DE by 26.1%. No site baseline or enrollment characteristics (including location or live training session attendance) correlated with measurable 3DE pairs; however, a low query rate during the quality process did correlate (p = 0.002).
Overall, the median quarterly 3DE measurability score was 69% (25th, 75th: 50%, 74%) and was also not related to site baseline characteristics. Although site 3DE measurability correlated with high subject enrollment (p = 0.031), it was more closely related to quality process characteristics (low query rate [p < 0.0001], no retraining needed [p = 0.010], and less time to certification [p = 0.047]).
The use of echocardiographic data for clinical research endpoints can be limited by poor image quality (4). We prospectively designed and tested a quality program to optimize echocardiographic quality in a clinical trial with 3 important findings. First, our program was feasible and yielded analyzable echo data in all subjects in whom imaging was attempted. Second, we achieved a very high rate of analyzable 3DE pairs (74%). Third, no pre-trial characteristics correlated with site performance. Compellingly, characteristics identified through the quality program correlated with site performance. These findings demonstrate the robust capabilities of echocardiography, the importance of ECLs, and their potential contributions to rigorous multisite research.
The quality initiative included attention to structure, process, and outcomes. As with other studies (3), aspects of “structure” included an imaging manual of operations and a site-certification process. Additional requirements not previously described included post-training quiz, individual sonographer certification, and monthly echo quality updates. Our “process” steps included previously reported quarterly scorecard feedback to sites but also novel requirements such as site physician approval of 3DE acquisitions, echo-specific investigator meetings with hands-on training, and retraining of underperformers throughout the study. Our “outcomes” included prospectively defining quantitative quality outcomes of 3DE pairs and measurability and identifying their correlates. As a result, paired echo data were obtained in all subjects, a far higher proportion than the previously reported 43.5% for analyzable baseline biplane 2DE (3). These results have important ramifications for use of echo endpoints in clinical trials and also suggest achievable targets for measurable 3DE data.
To our knowledge, this is the first study to examine site characteristics and processes that correlated with high-quality echocardiographic data in a randomized, multicenter, international clinical trial. Unfortunately, no baseline site characteristics predicted site performance, and only site characteristics identified through the quality program correlated with the primary and secondary quality endpoints.
Although randomization of sites to quality initiatives versus “usual” practices was not feasible, it is reasonable to assume that without intervention, the proportion of measurable echocardiographic pairs (baseline + 6 months) would have been no higher than the 47.9% of 3DEs certified on the first submission attempt. This is substantially different from the final result and quantifies the impact of our rigorous quality program.
Echocardiography laboratory quality initiatives can optimize site image quality as well as analytic rigor in clinical trials. These findings likely extend to the clinical setting.
Please note: This work was supported in part by the ASE Education and Research Foundation Award Number 12-G-10-ASE (ALC) and Bellerophon Therapeutics. The content is solely the responsibility of the authors and does not necessarily represent the official views of the American Society of Echocardiography or the ASE Education and Research Foundation or Bellerophon Therapeutics. All authors have received funding from Bellerophon Therapeutics for this project.
This project was an Internal Review Board (IRB)-exempt study on quality improvement for the Preservation 1 trial (NCT01226563).
- Crowley A.L.,
- Yow E.,
- Barnhart H.X.,
- et al.
- Chung E.S.,
- Leon A.R.,
- Tavazzi L.,
- et al.
- Rao S.V.,
- Zeymer U.,
- Douglas P.S.,
- et al.