While carotid (neck) artery imaging can be very helpful to evaluate patients who have had a stroke or mini-stroke, carotid imaging is commonly used on asymptomatic patients, for whom there is little evidence of benefit.
The Choosing Wisely campaign, launched in 2012, motivated many specialty groups to make reducing low-value carotid artery imaging a priority. In 2013, the Society of Thoracic Surgeons recommended against carotid imaging before surgery and the American Academy of Neurology recommended avoiding carotid imaging for healthy patients who fainted. In 2015 the Society for Vascular Surgeons cautioned against routine carotid ultrasounds for healthy, asymptomatic patients with carotid bruit (a murmur picked up on a stethoscope).
To what extent have efforts to change clinician habits worked, when it comes to low-value carotid imaging? A recent study in JAMA Network Open that examined the prevalence of low-value carotid imaging in the Veteran’s Health Administration over the course of a decade shows that we still have a long way to go. Carotid imaging for bruit declined from 2007 to 2012 (before Choosing Wisely launched), but the authors found no significant trends in rates of carotid imaging before surgery or carotid imaging for syncope (fainting).
To learn more, we spoke with Dr. Timothy Anderson, an Instructor of Medicine at Harvard Medical School and general internist at Beth Israel Deaconess Medical Center, and co-author of the JAMA Network Open study.
Lown Institute: What was the impetus for this study?
Dr. Timothy Anderson: We were interested in studying the impact of the Choosing Wisely campaign. Prior studies have looked at the impact just in the few months or years after the campaign launched, but we wanted to look at the longer-term impact. Clinician practice rarely changes quickly, so taking a broader look over ten years can give us a different view on the effect of Choosing Wisely.
We also wanted to use data from the VA (Veteran’s Affairs) system because we could access clinical information, not just billing records. The clinical data lets us see what doctors said they were ordering tests for, rather than relying on billing codes. Also, the VA is not a fee-for-service system, so we thought it was more likely that efforts to reduce low-value care would be working there, since there is less financial incentive for doctors to do this test. The VA is also a place where a lot of junior doctors are trained, so it’s a big part of medical education.
Why is it important that we reduce low-value carotid imaging in particular?
The cost of the test and the anxiety from false positives are reasons we commonly give for avoiding low-value testing. But cascade procedures are also a big concern. If you do carotid imaging and see blockages, one next step may be to consider surgery. These surgeries are invasive, you’re opening up the neck and stenting or scraping off plaque. You have a risk of stroke, death, and bleeding during these surgeries. And for people without a prior stroke, it’s less clear that the benefits of these surgeries outweigh the risks of the operation.
When you look at why asymptomatic people get these surgeries in the first place, it often happens after a test was ordered that was low-value. In our study, we found that 18.3% of carotid procedures that happened during the time period were preceded by low-value imaging.
Why do you think doctors continue to order these tests?
We didn’t look at this specifically in the study. But one reason could be that because carotid imaging is perceived as low risk. The test most often uses an ultrasound, there’s no radiation, so some doctors may see the test as harmless. There is a prevalent idea that if the test itself isn’t invasive, there’s no risk to it.
You and your co-authors used a different approach to identifying low-value testing. Can you describe how you did this?
Usually researchers use diagnostic codes from billing claims to find low-value care. But there are some downsides to this method–-sometimes the billing doesn’t happen until after the result of the test, or sometimes the bill doesn’t include enough details to determine the reason for the test. An advantage in using data from the VA is we could access the reasons for testing written by the clinicians ordering the test using the electronic health record. We developed a text-lexicon search and looked at imaging reports to group tests into categories of low-value, known stroke, and other.
On the Lown Hospitals Index, we found that overuse of carotid imaging for syncope was very common, but some of the best hospitals for avoiding this type of overuse were elite hospitals like the Cleveland Clinic, Yale-New Haven Hospital, Mass General, etc. Any thoughts on why that could be?
For any specific syndrome like fainting, it’s hard to say that an individual test is low value; it depends on the clinical context. But setting an institutional standard for patients with certain symptoms may help reduce low-value care by reducing variation in testing. Some hospitals have specific “pathways” for patients who come to the hospital with syncope or chest pain. That means they have set steps for what to do, strict algorithms, so these hospitals may be more likely to conform to high and low-value care guidelines. Often bigger academic hospitals have the resources to invest in these types of care pathways.
What can we do policy-wise to reduce this type of overuse?
We know that one effective mechanism to reduce low-value testing is not to pay for it, but this may not be possible in a fee-for-service system. Additionally, many tests are not always high- or low-value, but the value depends on the clinical context. Nonetheless, direct financial incentives are one way to change behavior. Prior authorizations, although they can be a source of frustration for doctors and patients, may be effective in reducing low-value care when used appropriately. Also, campaigns to educate and raise awareness like Choosing Wisely may have a more generational impact that our study did not capture; students who are training now may be doing things differently.
It’s certainly hard to change physician behavior. Often one size does not fit all, but this area of what I like to call “de-implementation science” is the key to improve the value of the care we provide.