The Importance of Being Expert: Is It Time to Revisit the Concept?

Experience matters in medicine. In the current era of lengthy training programs and sub-subspecialization, this statement is hard to dispute. The concept that experience matters is not limited to medicine, having also permeated popular culture. Malcolm Gladwell’s book Outliers , published in 2008, stresses the importance of repetition in achieving excellence, citing a pioneering study by Ericsson et al. demonstrating the relationship between hours of practice and skill level among musicians.

Few would argue that experience is not important. Yet how do we define what constitutes enough experience? There is significant evidence that high-volume operators perform procedures with fewer complications and better outcomes. However, the data on cognitive tasks are less well developed, and there is a paucity of data relative to echocardiography. In 1991, Picano et al. published a research study titled “Stress Echocardiography and the Human Factor: The Importance of Being Expert.” In that study, the investigators demonstrated that diagnostic accuracy in stress echocardiography improved by >20% with specific training. Now, 20 years later, Zhang et al. have published a study in this issue of the Journal of the American Society of Echocardiography examining the role of training in the tissue Doppler assessment of dyssynchrony.

Doppler tissue imaging (DTI), which is one type of strain imaging, measures tissue deformation. Strain imaging has been used to assess ventricular function, ischemia, and viability. Dyssynchrony analysis uses DTI to measure the contraction delay between ventricular walls. This technique showed initial promise in identifying individuals likely to respond to cardiac resynchronization therapy (CRT). However, when the results of the Predictors of Response to Cardiac Resynchronization Therapy (PROSPECT) trial were published in 2008, it quickly fell out of favor. In PROSPECT, which enrolled and randomized 498 patients at 53 centers, echocardiographic assessment of dyssynchrony showed only modest ability to predict outcome after CRT. Criticism of PROSPECT focused on the large interobserver and intraobserver variability in the trial (10%–15% and >30%, respectively), which suggested that DTI techniques were not yet ready for widespread implementation in clinical practice. It is not clear whether the application of the method, or the method itself, was flawed. This is an important difference as we struggle to find an accurate method to predict response to CRT.

Although previous studies of dyssynchrony analysis have focused on clinical outcomes, Zhang et al. chose to focus on the reproducibility of the technique. Specifically, the authors evaluated the ability of a hands-on training program to improve the reproducibility of dyssynchrony measures in patients with heart failure. This is important because patients with heart failure, specifically those with low ejection fractions and QRS widening, are the population in which CRT is used and therefore the most appropriate population in which to evaluate this technique. In this study, four cardiologists who were trained in echocardiography but did not have specific training in dyssynchrony analysis were randomized into two groups. One group received a 1-hour lecture on the technique followed by a demonstration of the methods and practical tips. The other group attended a 2-day structured training program. The authors note that this program consisted of both lectures and hands-on sessions in which the participants worked with trained tutors. Of note, the participants were required to complete 50 cases as part of the training program.

After completion of the training, the participants (called “beginners” and “graduates” by Zhang et al. ) performed dyssynchrony analysis on images from 70 patients. Importantly, images of varying difficulty were included in the study. The results of analyses performed by the beginners and graduates were compared with those of a cardiologist experienced in DTI (the “expert”). Four measures were obtained: the standard deviation of the time to peak myocardial systolic velocity among 12 left ventricular segments, the maximal difference of time to peak myocardial systolic velocity among the segments, the maximal opposite wall delay among the segments, and the maximal opposite wall delay among four basal segments.

The reference standard in this study was the results of the expert, which were compared with those of another cardiologist experienced in DTI. Initial analyses by both experts were obtained in a blinded fashion. If these results differed by ≥5 msec, the experts repeated the measurement together to come to a final decision. Determination of whether significant dyssynchrony was present was made as per American Society of Echocardiography (ASE) guidelines. The correlation coefficients for the beginners, compared with the experts, ranged from 0.292 to 0.643 for the four dyssynchrony indices that were assessed. This demonstrates, at best, a weakly to moderately positive correlation between measures made by beginners and experts. The authors also found a difference of ≥5 msec in nearly half of measures of standard deviation of the time to peak myocardial systolic velocity among 12 left ventricular segments performed by the beginners compared with the experts. Finally, the intraobserver variability for both beginners was >10%.

In light of these disheartening, but not surprising, results, it is a relief that the two cardiologists who had undergone a 2-day dedicated training program (the graduates) performed significantly better, with correlation coefficients ranging from 0.800 to 0.937, compared with the experts. Measures of both intraobserver and interobserver variability were lower in this group than in the beginner group. Zhang et al. also demonstrate that the trained readers were successfully able to interpret images of varying difficulty, although the variability increased with increasing difficulty level.

Although the principle that training leads to improved performance is not novel, it is nonetheless an important concept that is well highlighted by Zhang et al. The strengths of their study include the use of trained echocardiographers as the image interpreters in all groups (as opposed to trainees) as well as a “real-life” cohort of images from patients with abnormal left ventricular function. In addition, participants interpreted images of varying quality, reflecting the realities of clinical practice. The training program described by Zhang et al. is intriguing, although the authors provide little detail regarding the design and content of the training program. It is therefore hard to assess whether such a training program could be implemented at other centers.

Several weaknesses of this study deserve mention. This study was small, with only two cardiologists in each group. We do not know how much echocardiographic experience the participants had or whether there were any important differences between the two groups of nonexperts. However, the results of this study are intuitive, and it seems likely that echocardiographers of all levels of experience would show improved performance with training. It is not possible to know whether all echocardiographers would derive the same level of benefit as seen in this study.

There is no true gold standard for dyssynchrony analysis, and therefore there is no standard against which to compare the experts in this study. Although Zhang et al. used two expert readers to try to minimize this problem, an alternate study design would have used experts from different institutions, who might have different approaches to DTI. One can also argue that if the experts also taught the training course, they may have simply trained study participants to perform DTI in a style replicating their own, as opposed to actually improving their technique. Finally, in this study, DTI was used to assess dyssynchrony. DTI is increasingly being supplanted by other forms of strain imaging, such as speckle tracking. We can surmise that the results of this study also apply to other imaging modalities, including speckle tracking. However, we cannot make generalizations about how much training is required to become “expert” in those modalities without studies specifically addressing this question. We can say only that a 2-day, intensive training course appears adequate to perform dyssynchrony analysis by DTI.

Zhang et al. address in passing, but not in detail, the importance of image acquisition. DTI is angle dependent, and therefore the quality of the end product is dictated by the quality of the image acquisition. This also applies to other techniques, such as speckle tracking, that are not angle dependent but are reliant on adequate image quality. As echocardiographers adopt new technologies, they will need to focus not only on the interpretation but also on the quality of the image acquisition.

Two recent studies in this journal support the findings of Zhang et al. Johnson et al. demonstrated a significant improvement in the assessment of diastolic function after the implementation of a quality improvement project involving both physician and sonographer education. Tsang et al. examined the impact of experience on the interpretation of segmental mitral valve anatomy and found that expert readers (ASE level 3, >500 mitral valve cases) were significantly more accurate than intermediate (ASE level 3, <500 mitral valve cases) and novice (ASE level 2) readers. Analogous data exist in the radiology literature underscoring the importance of experience in the interpretation of imaging studies.

These results should come as no surprise to individuals who are involved in teaching fellows and trainees. There is a steep learning curve in the interpretation of echocardiograms. Although the introduction of novel techniques, such as strain imaging and three-dimensional echocardiography, provides an exciting opportunity for our field, these techniques are of little use if they are not performed in an optimal manner. This applies not only to clinical practice but also to the research setting, where issues of training and interobserver variability can have an important impact on trial results. Conversely, imaging techniques may appear promising in a research setting or core laboratory but be of limited use to untrained operators in the community. As more novel technology makes its way to clinical ultrasound machines, required validation includes not only assessment of the method but the optimal training program for applying the technology. Both manufacturers and professional societies have undertaken to close this gap. Zhang et al. ‘s study is important in that it sets a standard that should be considered by industry and educators. A lunchtime presentation is certainly not enough to learn the ins and outs of much of the newer software.

How should the results of this study change our practice? In Outliers , Gladwell alludes to the “10,000-hour rule,” which states that the key to achieving excellence in any discipline is to practice for 10,000 hours. Although the participants in Zhang et al. ‘s study achieved nowhere near this amount of practice with DTI, the basic principle still remains: rigorous training matters.

After the publication of Picano et al. ‘s study, the ASE adopted guidelines specifying the number of stress echocardiograms an individual needs to read to achieve proficiency in interpreting the technique. The results of the study by Zhang et al. imply that similar standards should apply to dyssynchrony analyses. As we continue to adopt new technologies, there remains a need for echocardiography, as a profession, to define and maintain standards of training, lest we forget the importance of being expert.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Jun 11, 2018 | Posted by in CARDIOLOGY | Comments Off on The Importance of Being Expert: Is It Time to Revisit the Concept?

Full access? Get Clinical Tree

Get Clinical Tree app for offline access