EVendo Score for Esophageal Varices

Why did you develop the EVendo Score for Esophageal Varices? Was there a particular clinical experience or patient encounter that inspired you to create this tool for clinicians?

We decided to embark on developing a score to predict presence and size of varices after reflecting on our experience as clinicians and recognizing the following:

a) many patients with cirrhosis who undergo esophagogastroduodenoscopy (EGD) screening for esophageal varices (EVs) are found to have no or only small EVs,

b) low-yield EV screening EGDs represent a potentially deferrable procedure that increases patient risk (from the procedure and associated anesthesia) and healthcare costs,

c) such EGDs take away resources from patients who have varices needing treatment (VNT) and are thus most likely to benefit from EGD, and

d) currently available predictive scores are limited in their applicability and can be onerous to use in clinical practice.

We realized there was a critical need to develop a means of better identifying the population in greatest need of EV screening and to reduce the burden of unneeded EGDs, and therefore we wanted to develop something novel that would be quick and convenient to use, yet highly accurate—hence the inception of efforts which led to the EVendo score.

What pearls, pitfalls and/or tips do you have for users of the EVendo Score for Esophageal Varices? Do you know of cases when it has been applied, interpreted, or used inappropriately?

The EVendo score was the result of extensive analyses of multicenter data via biostatistical techniques (in particular, random forest algorithm of machine learning, a powerful and innovative analytical method) and expert opinion, and it was validated for its intended use. However, even the most precise predictive scores can be subject to error or inappropriate use. Thus, while the EVendo score was developed and is intended to help with evidence-based clinical decisionmaking with regard to EV screening, it is not intended and should not be a substitute for objective clinical indicators and sound clinical judgment. Likewise, it should not be applied in contexts for which it was not validated (e.g. acute upper GI bleeding).

What recommendations do you have for doctors once they have applied the EVendo Score for Esophageal Varices? Are there any adjustments or updates you would make to the score based on new data or practice changes?

When the EVendo score is calculated for a patient and found to be low (≤3.90), it should be repeated periodically in accordance to the severity and nature of the patient’s cirrhosis/liver disease. We would generally advise recalculating on an annual basis, though adjustments may be made based on individual considerations and AASLD guidelines.

How do you use the EVendo Score for Esophageal Varices in your own clinical practice? Can you give an example of a scenario in which you use it?

I primarily like to use the EVendo score as a sieve to help avoid unnecessary EV screening EGD in patients who seem unlikely to have EVs based on clinical (including biochemical and imaging) features. For example, if I see a patient with compensated cirrhosis, normal CBC, and normal INR, my clinical intuition is that the patient is unlikely to have varices or will have only small varices, and the EVendo score can help me to determine this with more certainty. Thus, I will calculate an EVendo score, and if it comes out to ≤3.90, it confirms my a priori impression, and I recommend deferring the screening EGD.

I also like to use the EVendo score to help substantiate the need for EGD in patients who appear to have high periprocedural risk (e.g. cardiac comorbidities), realizing that the higher the EVendo score, the greater the likelihood of EVs (and thus the more benefit there may be from the EGD). Here are some positive predictive value (PPV) cutoffs that I consider:

• 10% PPV:  EVendo score of >4.57 and >5.38 for EVs and VNT, respectively.

• 25% PPV:  EVendo score of > 6.00 and >9.80 for EVs and VNT, respectively.

• 50% PPV: EVendo score of >15.15 for EVs.

If a patient has an EVendo score of, say, 10.2, I would strongly recommend an EGD for EV screening following appropriate pre-procedural (e.g. anesthesia) evaluation, realizing the PPV for having VNT would be at least 25%.

Any other research in the pipeline that you’re particularly excited about?

It is exciting to see more and more efforts and methods to provide tailored patient care which reduces healthcare burden in its various forms and gets away from historically more resource-intensive, non-individualized approaches based on scant data. This trend toward cost-conscious, evidence-based care is important and driven by many factors, and as biomedical researchers and clinicians, we must be engaged to help it take a form which keeps patient health and wellness at its center.

James H. Tabibian, MD, PhD, is an associate professor of medicine at the Geffen School of Medicine and the principal investigator of the Tabibian Laboratory at the University of California, Los Angeles. He is also an advanced endoscopist and the director of endoscopy at Olive View-UCLA Medical Center. Dr. Tabibian’s primary research is focused on advanced endoscopy and quality improvement.