Anyone who has spent time on the Internet has seen those annoying pop-up ads that contaminate commercial websites. Now, thanks to the Electronic Medical Record, physicians can enjoy the same type of interference as they are trying to negotiate patient charts.
One of the features of most EMR systems are pop-up alerts that trigger for any number of reasons and interrupt the flow of charting to inform you of some problem or demand some action. Some of these are useful, such as a pop-up warning that the patient is allergic to the antibiotic that you just tried to order. Others are merely annoying, such as the Medicare two midnight certification you have to attest to before you can make your over-65 patient an inpatient admission. But the net effect of all of these pop-ups is alert fatigue. They happen so often, and are so often for trivial reasons, that you tend to ignore them and click the boxes until the alert goes away and you can get on with your work.
I recently had six Priority Alerts (oh my!) trigger on one patient in the course of a routine 23hr postoperative observation period following an uncomplicated hernia repair. The patient was doing fine and none of the ‘Priority Alerts’ was valid, or in my opinion a priority. Two were completely off base and had I just ‘clicked boxes’ would have resulted in unnecessary imaging and testing with some potential morbidity.
How did we get here? I wish I knew. I tried to research some of the alerts through my department committee and went down a rabbit hole of diffused responsibility, well-intentioned effort and contradictory responsibility. These alerts can be placed in the system by just about anyone. The hospital administration places them to try to comply with CMS admission guidelines. Various departmental committees place them to improve compliance with best practice guidelines. The pharmacy is sometimes the worst offender, placing not only allergy alerts but drug interaction alerts that range from important, such as warning that two drugs have potentially fatal interactions, to the stupid, such as warning me that hypertonic solutions may cause phlebitis. Each of these may seem like a good idea, but they are given equal weight in the EMR (IT people and the computer can’t tell if something is important or trivial) so that the fatal interaction alert looks the same as the warning that ibuprofen should be given with food.
I asked some of my hospitalist friends what they do to handle these alerts and most of them have gotten to the point where they ignore them. So if most doctors are ignoring the alerts, what good are they?
There are a few good studies of specific types of alerts such as severe drug interaction warnings or allergy warnings that seem like a good idea but failed to show significant improvement in the incidence of these types of errors. Why? I suspect because we were already pretty good at looking for this type of stuff with multiple layers of check by the physician, the pharmacy and the nursing staff.
Other alerts, such as the sepsis alert that triggers for certain combinations of abnormal vital signs, seem to show better results at the institutions where they were tested, but no one has looked at the numbers of patients for whom the interventions triggered were not necessary. And a careful look at the data shows that many of these institutions had poorer than average results BEFORE the alert was instituted. It isn’t as hard to make a poor outcome look better as it is to make a good outcome look stellar.
The only solution I can see is to create yet another layer of bureaucracy to oversee the inclusion of automatic alerts into the record. That sort of solution grates on me, though. We have too many committees and advisory groups telling us how to practice medicine as it is. Creating some physician oversight of these alerts would return some measure of control to the practicing doctor, but only at the cost of more time spent in a meeting room and less time at the bedside.