Eyeballing of Lesion Lengths Often Inaccurate, Can Impair Stent Choice

Interventionalists’ visual estimation of coronary lesion length is highly variable and results in overstenting or understenting in about one-quarter of cases, according to a study published online January 23, 2015, ahead of print in Catheterization and Cardiovascular Interventions. Take Home: Catheterization and Cardiovascular Interventions

Calling the results “unnerving,” John A. Spertus, MD, MPH, of St. Luke’s Mid America Heart Institute (Kansas City, MO), told TCTMD in a telephone interview that the interventional community faces a “real challenge… to redesign care to make sure that they are most accurately treating patients [for] their actual disease.”

Researchers led by Paul T. Campbell, MD, of the Sanger Heart & Vascular Institute (Concord, NC), invited 40 interventional cardiologists to review 25 matched orthogonal angiographic images that were prescored using QCA. All images were of simple or complex single de novo lesions with a vessel diameter of 2.5-3.5 mm, pre-PCI lesion lengths of no more than 20 mm with length documented for each view, between 50% and 100% stenosed, and implanted with a single stent. Five of the cases were repeat images to test intraoperator variability.

The interventionalists were mostly between 40 and 59 years old (78%), and the majority had been in practice for at least 11 years (80%). Most came from teaching hospitals (65%) and performed at least 100 PCIs annually (80%). Importantly, 95% of respondents said that 2-4 mm was their target stent overlap of the proximal and distal edges.

Minority of Estimates Matched QCA Measurements

Lesion lengths reported in the surveys were within -1 to + 4 mm of QCA measurements 30.4% of the time. About half of measurements (51.1%) were short, with 29.9% of lesions being underestimated by more than 4 mm. Lesion measurements were too long in 18.5%. Physicians who reviewed the images later in the day were less likely to report accurate measurements (P = .033 for trend).

Similar patterns were seen for stent selection. Only 22.3% of respondents chose proper stent lengths, defined as 2-4 mm overlap of the proximal and distal edges of the QCA-measured lesion. Stents were short of the optimal length in 55.0% and below the QCA-measured lesion length in 23.8%. Stents were longer than the optimal length (≥ 8 mm) in 22.8%.

Individual operators tended to give different answers regarding lesion length (P = .11) and stent selection (P = .67) when looking at repeat images of the same lesion. There was a difference of greater than 3 mm in 38.5% of lesion length measurements and 37.5% of stent length selections.

‘An Important Reminder’ That Stent Length Matters

“Visual assessment of the coronary lesion length has a high degree of interrater and intrarater variability, which may lead to inadequate lesion treatment and coverage or alternatively to overcoverage,” Dr. Campbell and colleagues write. “Employing methods to improve the accuracy of lesion measurement may reduce [incomplete stent coverage], overstenting, and improve patient outcomes.”

Dr. Spertus commented that both underestimation and overestimation of lesion length are issues, although common practice is to buffer the choice of stent length by a few extra millimeters. “You really want to match the stent length to the lesion,” he said. “Underappreciation of the length of the lesion can affect both your choice of a drug-eluting or bare-metal stent and inadequate coverage of the lesion, leading to increased risk of restenosis. And overestimating it can increase the cost of the procedure.”

Brahmajee K. Nallamothu, MD, MPH, of the University of Michigan (Ann Arbor, MI), said the study is valuable because estimation of lesion length has become something of a lost art. “In the drug-eluting stent era, most of us just say, ‘Why don't we go a little bit longer? No harm, no foul,’” he told TCTMD in a telephone interview. “But [the study] is correct that issues around stent length selection are not inconsequential, and trying to focus on this part of how interventionalists read angiograms is an important reminder for us.”

Generalizability Unclear

“These were sort of ideal lesions, carefully selected,” Dr. Spertus commented. “Routine clinical practice—where there's more angulation and maybe not full dilatation of the vessel with nitroglycerin—is probably worse than this.” Since patients are not often able to choose their interventionalists, he said, “it is incumbent on the profession to be sure that we're being very accurate and highly reliable in our estimates of a patient's disease and the optimal treatment.”

Dr. Nallamothu took issue with the lack of data on lesion characteristics, noting it is unclear where the images came from, what types of lesions were involved, and how appropriately QCA was used to determine length. “So the generalizability is definitely a challenge,” he said.

Also, the authors “don't tell us where these 40 interventional cardiologists came from,” he commented, adding it remains uncertain as to whether they all practice in the United States. “They talk about an online survey, but was this tied to an academic meeting or a statewide initiative or something [else]?” The answer is relevant to determining how widespread this problem is, he added.

Fixing the Problem

While it may be more efficient to see a lesion and instantly treat it, Dr. Spertus said, “a more accurate process of care may be to do a diagnostic angiogram and then pause. Take the time to do QCA or have a second cardiologist really reflect on what is the right lesion length.”

The next level would be “retraining and recalibrating,” Dr. Spertus observed, suggesting that physicians review lesions monthly to continually refine their skills. “If the profession doesn't respond to this and improve the accuracy and reliability of our estimates, then eventually we are opening the door for some regulatory agency to step in and do it. And I hope that doesn't happen,” he concluded.

According to the authors, data indicate that technologies that provide more objective measurements of lesion lengths, such as QCA, CT angiography, and IVUS, can improve outcomes.


Campbell PT, Mahmud E, Marshall JJ. Interoperator and intraoperator (in)accuracy of stent selection based on visual estimation. Catheter Cardiovasc Interv. 2015;Epub ahead of print.




  • Dr. Campbell reports serving on the medical advisory board of Corindus and as the principal investigator of the Precision Registry.
  • Drs. Nallamothu and Spertus report no relevant conflicts of interest.


Related Stories:



We Recommend