Commentary: How ‘Alert Fatigue’ Truly Exhausts Us

My organization required our physicians to complete continuing medical education online modules about provider burnout. Based on a national concern of doctors becoming burned out from the dehumanizing burden of regulatory and administrative requirements and the use of electronic medical records in modern medicine, we must complete obligatory computer modules. But I digress.

One module taught about “combating alert fatigue.” I took note that “providers using an EMR system may encounter more than 100 alerts daily.” Seemed about right. It went on to explain one “strategy to prevent alert fatigue is for an internist to customize his/her alerts to try to minimize the number of unnecessary alerts, allowing an internist to focus on fewer alerts.”1

The author of the module, if he or she is indeed a provider, must use a different EMR system than I use, as there is no customization available for our alerts. Furthermore, it feels as if new alerts are added daily. I suppose the documentation requirements for compliance, the ICD-10 upgrade, MIPS, and prior authorizations for medications and devices just hasn’t slowed our clinical care enough. Bring on the alerts!

Did you know the National Library of Medicine now has a Medical Subject Heading titled “Alert Fatigue, Health Personnel?” It is a lower branch off several other MeSHs: “mental fatigue,” “attitude of health personnel” and “quality assurance, healthcare.”

So, I performed a PubMed search using “Alert Fatigue, Health Personnel” that revealed only 48 publications on this subject. Are you surprised? When I restricted the search to “Clinical Trials,” I found only one clinical trial in the group which, strangely enough, was about public health messaging to providers by email, fax and short message service — not EMR alerts. In this study, the authors found every “increase of one local public health message per week resulted in a statistically significant 41.2 percent decrease … in the odds of [providers] recalling the content of the study message.”2

An increase of one public health message per week. Imagine the proportional impact on providers when EMR alerts fire at a rate of 100 every eight hours. (My bet is they suck out all knowledge from the human brain.)

Reviewing the other 47 abstracts in my search, I grew concerned about the science behind these electronic invaders of our day: No clear definition or agreed upon metrics exist for “alarm fatigue,”3 clinical decision supports often malfunction,4 and alert desensitization has led to sentinel events and death.5 One interesting retrospective cohort study showed that the likelihood of a provider’s acceptance of an alert “dropped by 30 percent for each additional reminder received per encounter.”6

The most revealing and realistic assessment of human behavior in these articles posited “alert override becomes habitual. The override response is activated by environmental cues and repeated automatically, with limited conscious intention.” In other words, when providers see an alert — much like smokers and the Surgeon General’s warning — they no longer think about it.

We rail against device manufacturers and pharmaceutical companies for lack of due diligence in rigorous scientific trials. But who is calling for randomized trials of EMR alerts and clinical decision supports? These taps on the shoulder, these nudges, hold the promise of improving our patient care. Intuitively, they might even make sense. But when you unleash a platoon of these discourteous bots into a clinical encounter, the provider must decide: “Should I take the requisite time required to read the alert in detail? Or should I blow past them at 78 mph in order to complete my patient encounter and move on to the next?”

Recently, a patient had a terrible toothache and called because she couldn’t get to a dentist. Could I send some antibiotics? I entered my prescription for amoxicillin, and a pop-up appeared. I tried to go around it — it countered. I tried a basketball move: a head-fake and a spin. It thwarted me. With clenched teeth, I clicked on the little beast. “Patient had a past severe reaction to penicillin.” Curses! I almost had this harangue completed when I had to admit what I already knew, deep down: Alerts have their place.

But I do believe they, like mosquitoes in summer, are making it unpleasant to complete the required work. We need sanity. Having each proprietary EMR with its own set of alerts that are untried, untested, based upon good ideas rather than on good science, and burgeoning in number and force is marching the clinical enterprise through a morass.

Once alerts are studied and proven effective, once we know the number and timing of alerts that are critical to ensure safe patient care, once we see data demonstrating their efficacy — “Number Needed to Alert” to save a life or prevent harm — then all EMRs should install them and have them presented to providers the same way, regardless of the vendor.

I don’t think we are asking for much here. We just want to be left alone — most of the time anyway. We want alerts that make it easier to do the right thing for patients without being impeded.

References

1. Med -IQ, Inc. Combatting Alert Fatigue. In Risk and Safety Issues in Internal Medicine. Retrieved from med-iq.com/access-education on May 17, 2018.

2. Baseman JG, Revere D, et al. Public health communications and alert fatigue. BMC Health Serv Res. 5;13:295, Aug. 5, 2013.

3. Winters BD, Cvach MM, et al. Technological distractions (part 2): a summary of approaches to manage clinical alarms with intent to reduce alarm fatigue. Crit Care Med. 46(1):130-7, 2018.

4. Kassakian SZ, Yackel TR, et al. Clinical decisions support malfunctions in a commercial electronic health record. Apl Clin Inform. 8(3);910-3, Sept. 6, 2013.

5. Johnson KR, Hagadorn JL, Sink DW. Alarm safety and alarm fatigue. Clin Perinatol. 44(3):713-28, Sept. 2017.

6. Ancker JS, Edwards A, et al, Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Med Inform Decis Mak. 17(1):36, 2017.