NUR4083 Module 4 clinical decision support analysis example

An Alert on Seventy-Eight Percent of Admissions: Why a Fall Risk Advisory Stopped Working and How to Make It Mean Something Again

Student Name

American College of Education

NUR4083: Nursing Informatics

Module 4 Assignment

Instructor Name

February 1, 2027

The Alert and Its Numbers

The hospital in this paper, a composite built for the assignment, added an interruptive advisory to its electronic record three years ago after a serious fall. The advisory appears when a nurse opens the admission assessment for any patient who meets at least one of five criteria: age 65 or older, any sedating medication ordered, a history of falls in the past year, a fall risk score in the high range or impaired mobility documented by the emergency department. The pop-up window states that the patient is at high risk for falls and asks the nurse to select one of three responses: fall precautions initiated, fall precautions already in place or dismiss.

A review of the most recent quarter of data by the informatics team found that the advisory fired on 78 percent of admissions to adult medical-surgical units, 3,914 of 5,018. Nurses selected dismiss or fall precautions already in place, without opening the care plan, on 91 percent of those firings, and the median time the advisory stayed open was 1.8 seconds. During the same quarter, the hospital's fall rate on those units was unchanged from the year before the advisory was introduced. An alert that appears for more than three-quarters of patients has stopped telling nurses anything they did not already assume.

Why Alerts Lose Their Force

The pattern in these numbers is well documented. A systematic review of drug safety alerts in computerized prescriber order entry found that clinicians overrode alerts in 49 to 96 percent of cases (van der Sijs et al., 2006). The authors identified error-producing conditions in the alerting systems themselves: low specificity, low sensitivity, unclear information, unnecessary workflow disruption and inefficient handling. They argued that improving safety requires fixing those conditions rather than simply urging clinicians to pay more attention. The fall advisory has every one of those conditions. Its criteria are so broad that most patients meet them, it adds nothing beyond what the nurse is already documenting, and it interrupts the admission assessment at its busiest moment.

Repetition makes the problem worse. Ancker et al. (2017) studied alerts and reminders in primary care and reported that reminder acceptance fell as the number of reminders grew, with the likelihood of accepting a reminder dropping by 30 percent for each additional reminder in an encounter and falling further as the proportion of repeated reminders rose. Nurses on this unit see the fall advisory on most of the patients they admit, every shift, which is the pattern that research associates with declining response. The problem is not that nurses do not care about falls; it is that the alert has taught them that it is almost always noise.

The Five Rights Applied

Osheroff et al. (2012) describe the five rights of clinical decision support: the right information, to the right person, in the right intervention format, through the right channel, at the right point in the workflow. Applying them to the fall advisory shows where it fails. The information is not right, because it restates risk factors the nurse is already recording and says nothing about what precautions this particular patient needs. The person is arguably right, since nurses implement fall precautions, but the alert does not reach the nursing assistants who carry out most of them.

The format is wrong for most patients. An interruptive pop-up is appropriate for a rare, high-stakes situation that needs an immediate decision, and fall risk in hospitalized older adults is neither rare nor something decided in a second. The channel, a window inside the admission form, isolates the message from the care plan where precautions are actually ordered. And the timing is wrong: the alert fires when the nurse opens the admission assessment, before she has assessed the patient, so it asks for a decision before the information needed to make it exists. Each of the five rights is missed at least partly, which explains the override rate better than any claim about nurse behavior could.

A Redesign

The redesign replaces the single interruptive advisory with two tiers. For most patients who meet any fall risk criterion, the record will stop interrupting and instead add fall precautions automatically as a suggested section of the nursing care plan, displayed alongside the fall risk score after the nurse completes it. That moves the information to the channel and time where precautions are actually ordered and removes the interruption for the majority of patients.

An interruptive alert will remain for a narrow group where the risk is high and a specific action is often missed: patients who score in the high range on the fall risk tool and have a new sedating medication given in the past four hours, or who fell during the current admission. For that group, the alert will name the specific trigger, for example that lorazepam was given at 1400, and offer two concrete actions, activating the bed exit alarm and scheduling toileting every two hours, each of which can be ordered from the alert with one click. Based on the quarter's data, the informatics team estimates that the interruptive alert would fire on roughly 6 percent of admissions rather than 78 percent.

Measuring Whether the Redesign Works

The redesign will be judged by four measures over two quarters. The interruptive alert's firing rate should fall to under 10 percent of admissions. Its acceptance rate, defined as the proportion of firings in which the nurse orders at least one of the offered actions, should rise above 50 percent. The proportion of high-risk patients with fall precautions in the care plan within eight hours of admission should not fall below its current level, which confirms that removing the interruption did not remove the precautions. And the medical-surgical units will keep tracking their fall rate and their rate of falls with injury, each counted per 1,000 patient days, though the team accepts that falls have many causes and a change in one alert is unlikely to move them quickly.

The nurses who work with the alert will be part of the evaluation. Two staff nurses from each unit will review a sample of interruptive firings each month and report whether the alert was useful, which gives the informatics team information that override data alone cannot provide.

Conclusion

The fall risk advisory in this hospital fired on 78 percent of admissions and was dismissed in under two seconds almost every time. Research on alert fatigue explains why: alerts that are unspecific, repetitive and badly timed lose their meaning, and clinicians learn to ignore them. The five rights of clinical decision support show exactly where this alert fails and point to a redesign that removes the interruption for most patients and makes it specific and actionable for the few who need it. An alert earns attention by being rare and right, and a nurse's click on dismiss is often a judgment about the alert rather than about the patient.

References

Ancker, J. S., Edwards, A., Nosal, S., Hauser, D., Mauer, E., Kaushal, R., & HITEC Investigators. (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, Article 36. https://doi.org/10.1186/s12911-017-0430-8

Osheroff, J. A., Teich, J. M., Levick, D., Saldana, L., Velasco, F. T., Sittig, D. F., Rogers, K. M., & Jenders, R. A. (2012). Improving outcomes with clinical decision support: An implementer's guide (2nd ed.). HIMSS.

van der Sijs, H., Aarts, J., Vulto, A., & Berg, M. (2006). Overriding of drug safety alerts in computerized physician order entry. Journal of the American Medical Informatics Association, 13(2), 138-147. https://doi.org/10.1197/jamia.M1809