Q&A: Alert fatigue is one of CDS' shortcomings
 |
The result has been termed “alert fatigue;” when CDS systems give too many alerts and providers tend to ignore alerts altogether rather than sift through the less useful information. These and other flaws have discouraged many providers from using CDS systems at all.
Kesselheim wrote about alert fatigue with several colleagues in an analysis in the December issue of Health Affairs, concluding that a collaborative effort needs to occur between vendors, providers and the federal government to develop more effective CDS systems.
In an interview, Kesselheim spoke about the problem.
In their current form, what are the benefits of CDS alerts?
There’s an enormous amount of medical knowledge out there, more than any one person can retain at one time. When it’s working effectively, CDS systems can inform the choices physicians make by supplementing them with accurate data. They’re helpful in providing information about patients they’re treating and also about the treatment that they’re choosing.
If a person is on 12 medications and a physician is writing a thirteenth, the computer will automatically test that against the other 12 to make sure there aren’t going to be interactions that the physician should be aware of. Physicians are supposed to think about that, but there are limits to what people can hold in the front of their minds at one time, so these kinds of supplements can be very useful.
In their current form, what is the problem with CDS alerts?
Many CDS systems are not sufficiently helpful because they have too many inconsequential and irrelevant alerts built into them in an attempt to be as comprehensive as possible. The downside of being overly comprehensive is that, studies have shown, physicians will be more likely to tune out all of the alerts and then you’re back where you started.
The point is that alerts can be very helpful, but if there are too many, then physicians are behaviorally modified to skip all of the alerts.
Would more physician involvement in CDS system development improve their usefulness?
Individual physician involvement with vendors might be useful, but could also introduce too many variables. I think that rather than having individual physicians involved at the vendor level, there needs to be more give and take between the vendors and their large institutional customers, like academic medical centers, because those groups tend to know best what kinds of issues their physicians are dealing with and what they need support with.
Also, there needs to be more interaction between vendors and regulators, who can help provide guidance based on the totality of user experiences and the range of data regarding which CDS alerts have been helpful.
As the technology advances, will vendors begin developing more useful CDS systems on their own?
The technology isn’t that young. We’re doing all kinds of crazy things with computerized provider order entry [CPOE] and EHRs, so I don’t think that it’s too difficult to conceive trying to implement more useful CDS systems. I think it’s more a matter of culture–the culture of the vendors developing technology and the culture of the physicians using it.
If physicians realized how CDS alerts can help them, then they would be more accepting of the systems. If CDS alerts were developed in a more evidence-based way, for example by favoring the most important warnings to demonstrate maximum effectiveness, then more physicians would use CDS systems. It’s less a matter of the technology than it is about changing the culture and the expectations that physicians have for CDS alerts.
Different specialties have different needs. Does that explain the overabundance of alert? Even within specialties, is there a need for a wide range of products?
The amount of flexibility that physicians have in implementing CDS systems is relatively limited. There are substantial differences between the specialties and what they need from CDS systems is based on the treatment decisions they make. But as far as I can tell from watching physicians work in the field, varying CDS system designs for different specialties is not happening as often as it should.
I think that there’s probably a lesser range within a specialty, but there should still be options and opportunities for end-users to adapt CDS systems to their local environments in order to increase effectiveness. Who knows better what a large group of cardiologists in a particular location need better than they do?
In a December Health Affairs article, you and your co-authors make an argument for increased government involvement. How would that help?
If the government was more engaged in developing policies that define what is and what isn’t an effective CDS system, then the vendors themselves would feel less apprehensive about taking the steps necessary to create effective CDS systems.
Right now, vendors are worried in part about opening themselves up to liability. If vendors received some kind of validation that their CDS system development process met the government’s standards, then I think the reticence to take those steps would be reduced. When physicians prescribe drugs in accordance with federal guidelines, they’re generally protected from liability. Vendors might like the same protection.