Operationalizing Effective Alerts

Decision support offers the opportunity to provide smarter intelligence for our clinicians at the point of order entry. The Choosing Wisely initiative is guiding our efforts to more effectively order appropriate medications and studies.

Choosing Wisely, you may know, is a national effort to reduce the ordering of tests and medications that have not been shown to be helpful in specific populations and may actually  waste money and resources and lead to adverse patient outcomes.

If you’re really going to operationalize these rules suggested by a wide variety of specialty organizations, you have to automate the process. We evaluated  the current list of about 140 suggested rules (more are being added each month) and identified the high-priority items—the things we think would have the greatest impact on clinical quality and patient safety.

We have already built seven rules and are now in the process of deploying those rules in what we call background mode. The rule is checked and an alert fired but it’s not actually displayed to clinical users. Each alert is logged into our database and  our clinical informaticists will determine how frequently they fire, whether the logic built in is sufficient and the likely beneficial outcomes of implementation.

We have initially selected several rules related to imaging. Two relate to not ordering imaging studies for uncomplicated headaches or for low back pain within the first 6 weeks of onset since these can be expensive and the radiation is not counterbalanced by the minimal chance of finding something treatable. Another is not performing routine chest x-rays before surgery because patients who don’t have lung or chest symptoms almost never have findings on chest x-ray worthy of evaluation.

Another rule we prioritized is to discourage the use of valium or other related drugs in older adults for insomnia because it’s almost never necessary and can lead to confusion and falls.

Physicians are more likely to become immune to alerts when they see too many reminders about things they already know, even when they are appropriate. So many are not helpful that clinicians may become  accustomed to that, so that even when one is helpful, I could be conditioned to override it and move on.

To combat alert fatigue, our No. 1 priority in clinical decision support is to tell clinicians something they don’t already know. By and large, physicians already know what they’re doing but they might not have read the latest journal article on a related topic. Or, maybe they missed  seeing something relevant in the patient’s chart. That’s where we want to fire a reminder or warning to provide useful information. We know that still might not change the physician’s decision because, at least in 2014, no computer knows as much about one patient as the doctor taking care of the patient.
As with most initiatives related to health IT, this is a work in progress.

Harris Stutman, MD, is Executive Director of Clinical Informatics at MemorialCare Health System in California.