HI501 Unit 9 Seminar Reflection Example

I Came to Seminar Wanting a Stricter Threshold and Left Arguing About Who Should Get the Warning: A Unit 9 Reflection on a Nephrotoxin Alert

[Student Name]

Purdue University Global

HI501: Health Informatics

Unit 9 Seminar Reflection

[Instructor Name]

[Date]

The patient, the alert and its figures are composite; the guideline and studies cited are real.

The Case

For classmates who completed the written alternative, here is the case in five sentences. A 71-year-old man, a composite patient, was on day two after a hip repair when a resident ordered ketorolac for pain. Over the previous 36 hours, his serum creatinine had climbed by 0.3, to 1.2. The alert warning about a nephrotoxic medication with a recent creatinine rise fired, and the resident overrode it in four seconds with the reason "benefit outweighs risk." Two days later his creatinine was 2.1, and the ketorolac was stopped by the pharmacist on rounds.

Where I Started

I came to the session with a clear position. The alert fired on about 17 percent of nonsteroidal anti-inflammatory orders on our surgical units, so it had become noise, and residents had learned to dismiss it. My fix was to tighten the trigger from a creatinine rise of 0.3 to a rise of 0.5, which by my estimate would cut firings by more than half. Research on alert fatigue supported me: the more often clinicians see repeated alerts, the more likely they are to accept none of them (Ancker et al., 2017). I expected the class to agree.

What Happened in the Seminar

A classmate who works nights on a surgical unit took my proposal and applied it to the case. At the moment the ketorolac was ordered, the patient's rise was exactly 0.3. Under my stricter trigger, the alert would not have fired at all. She pointed out that a rise of 0.3 within 48 hours is the smallest change the international guideline counts as acute kidney injury (Kidney Disease: Improving Global Outcomes Acute Kidney Injury Work Group, 2012), and that early, small rises are exactly when stopping a nephrotoxic drug does the most good. My fix would have silenced the alert for the very patient in front of us. Then the instructor asked something none of us had considered: what does the override actually document? I answered that it documents the resident's judgment. The instructor asked us to look again. The override reason, "benefit outweighs risk," was the first option in the list and was chosen in four seconds. It documented only that the alert had been seen. The alert was not failing because it fired too often; it was failing because it fired at a prescriber at the one moment when a prescriber is least able to stop and think about kidney function.

From a Numbers Problem to a Recipient Problem

I had treated alert fatigue as a numbers problem, too many alerts, and reached for the obvious lever, a higher threshold. The discussion turned it into a question of recipient and timing. At order signing, the resident is thinking about pain, not about a creatinine trend from yesterday. The person who reviews every active medication against the latest labs every day is the unit pharmacist. A principle of good decision support is that the right information should reach the right person at the right point in the workflow (Bates et al., 2003), and I had kept the alert pointed at the wrong person while trying to make it quieter. The group's revised design kept the 0.3 threshold but changed the recipient. A creatinine rise of 0.3 or more in a patient on a nephrotoxic drug creates a task in the pharmacist's queue within the hour, with the trend and the drug shown together. The prescriber still sees an alert at signing only when kidney injury has already been documented. Firings to prescribers fall, sensitivity stays and the review happens when someone has time to do it. We also agreed that the override list itself needed work. If an alert still interrupts a prescriber, its reasons should be specific enough to mean something later, such as "creatinine rise explained by contrast, rechecking tomorrow" or "short course, kidney function monitored daily," and the first option in the list should not be the vaguest one. A pharmacist reviewing overrides at the end of the month could then tell judgment from reflex.

Three Questions Before Any Threshold

In my own work on decision support, I will ask three questions before I propose changing any threshold: whose action is this alert asking for, when is that person best placed to take it, and what does the override record? I will also apply any proposed rule to the case that started the discussion before I present it, which is what my classmate did in about two minutes and I had not done at all.

References

Ancker, J. S., Edwards, A., Nosal, S., Hauser, D., Mauer, E., & Kaushal, R. (2017). Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Medical Informatics and Decision Making, 17(1), Article 36. https://doi.org/10.1186/s12911-017-0430-8

Bates, D. W., Kuperman, G. J., Wang, S., Gandhi, T., Kittler, A., Volk, L., Spurr, C., Khorasani, R., Tanasijevic, M., & Middleton, B. (2003). Ten commandments for effective clinical decision support: Making the practice of evidence-based medicine a reality. Journal of the American Medical Informatics Association, 10(6), 523-530. https://doi.org/10.1197/jamia.M1370

Kidney Disease: Improving Global Outcomes Acute Kidney Injury Work Group. (2012). KDIGO clinical practice guideline for acute kidney injury. Kidney International Supplements, 2(1), 1-138.