In a Letter to the Editor, published in Seminars in Thoracic and Cardiovascular Surgery¹, Transonic points out a number of anomalies in the statistical analysis of a study by Kim et al.², and calls for more collaboration and open research.

This study aimed to answer two highly relevant questions:

1. Do threshold values for transit time flow measurement (TTFM) metrics in coronary artery bypass graft (CABG) patency assessment need adjustment for venous sequential grafts?
2. Do the metrics pulsatility index (PI) and mean flow rate (Q_mean) have predictive value for long-term failure of sequential venous grafts, and for major adverse cardiac and cerebrovascular events (MACCEs)?

In determining threshold values for Q_mean, the statistical analysis expressed the assumption that not low, but high Q_mean indicates graft failure. From other information in the paper, and from the author's reply to Transonic's letter³, it is clear that the authors are well aware that this is not the case. Still, the resulting threshold values for Q_mean are likely incorrect, and the conclusion that PI is a significantly better indicator of sequential graft failure than Q_mean cannot be drawn based on this study.

More importantly, despite the impressive number of 1288 sequential venous grafts included, the study size is insufficient for multivariable logistic analysis, as only 37 MACCE cases were found. This attests to the great skill of the surgeons involved, but it also means that the maximum number of variables to include in Cox regression is limited to 2 or 3 (each variable requires at least 10-20 events). The list of known confounders included in the paper is far longer than that, meaning that, despite the intelligence and effort that were given to this study, the conclusions drawn from this analysis are not statistically valid. In other words, the conclusion that PI is an independent risk factor for MACCE may not be drawn from this study.

To make any meaningful progress in assessing the long-term predictive value of TTFM, much larger, carefully designed studies are required. This may, for example, be achieved by setting up multi-center studies, or by systematically sharing the results of smaller studies - including a detailed description of measurement protocols - in public databases like PhysioNet^4,5.

Without pre-registration, fully describing measurement protocols and sharing raw results ("open research"), clinical studies risk the publication of invalid and misleading results, potentially leading to erroneous decisions in the operating room, and wasted time and resources in follow-up studies.

In their reply to Transonic's letter, the authors express their support of sharing results. Therefore, Transonic is hopeful that Kim and colleagues may be willing to collaborate in the future, and thus to become part of Transonic's long tradition of collaboration-based innovation.

References:

• Drost CJ and Drost S. Sequential Coronary Artery Bypass Grafts: Flow-based Patency Assessment Criteria. Seminars in Thoracic and Cardiovascular Surgery 2024.
• Kim H-H, Kim J-H, Lee SH, Yoo K-J, and Youn Y-N. Transit-Time Flow Measurement and Outcomes in Coronary Artery Bypass Graft Patients. Seminars in Thoracic and Cardiovascular Surgery 2023; 35(2).
• Kim H-H and Youn Y-N. Reply: What is the suitable predictor for venous sequential coronary grafting? Seminars in Thoracic and Cardiovascular Surgery 2024.
• https://www.physionet.org/ (last accessed 17 April 2024).
• Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov PCh, Mark RG, J.E. Mietus et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation, 101(23):e215–e220, 2020.