Who Owns the Alert?
A health system can deliver an alert perfectly and still fail if no one is responsible for what happens next
The scan had answered the question we asked of it. It had also found something we had not been looking for.
The patient was in his early seventies and had undergone CT imaging for a vascular problem that, fortunately, did not require immediate intervention. We were discussing that reassuring part of the result when his wife unfolded a sheet of paper she had brought from home.
“What about this?” she asked.
She pointed to a sentence farther down the radiology report.
There was an unexpected abnormality in the chest, suspicious enough that the radiologist had recommended further evaluation. The finding was not an emergency, but it was not something to forget.
I looked through the record.
The report had been finalized. An electronic alert had been generated. More than one clinician had received it. The result was visible in the patient portal. From an information technology perspective, very little had gone wrong.
Yet I could not immediately find the answer to the question that mattered most.
Who was taking responsibility for the finding?
The patient looked from the screen to me.
Test result follow-up is often described as a communication problem. I have come to think it is equally a problem of ownership.
“Was somebody going to call me?”
That question has stayed with me because it exposes a weakness that better detection alone cannot solve.
Modern medicine has become remarkably good at making abnormalities visible. Imaging finds incidental lesions. Laboratories flag unexpected values. Electronic records highlight overdue actions. Algorithms identify patterns. Remote systems can detect deterioration before a patient feels different.
We have spent enormous effort improving the signal.
We have been less disciplined about deciding who owns what happens after the signal arrives.
Notification is not ownership
An alert appears to be an action because something has moved.
A result has left the laboratory. A radiology report has entered the record. A notification has arrived in an inbox. Perhaps several inboxes. Someone may even have clicked on it.
But movement of information is not the same as movement of responsibility.
Researchers have been documenting this distinction for years. In one study of 1,196 critical imaging alerts in an integrated electronic medical record, 18 percent were not electronically acknowledged within two weeks, and 7.7 percent did not receive timely follow-up within four weeks. More surprisingly, alerts sent to two clinicians rather than one were associated with a greater risk of lacking timely follow-up. The safeguard of notifying more people appeared, in some cases, to make responsibility less clear. 1
That finding should make us uncomfortable.
We tend to assume redundancy produces safety. Often it does. But redundancy without explicit ownership can create another phenomenon: each person reasonably believes someone else may be acting.
The alert has recipients.
It does not necessarily have an owner.
This problem is not evidence that clinicians are careless. It exists partly because contemporary medicine produces an extraordinary volume of information. Primary care physicians in one study received an average of nearly 77 electronic notifications each working day, including test results, referrals, medication requests, and messages from patients and other clinicians. 2 More recent national data show that electronic message volume and the time physicians spend inside the electronic health record remain substantially higher than before the pandemic. 3
When every item asks for attention, the difficult work is no longer simply seeing information.
It is distinguishing what requires action, deciding who should act, and making sure the action actually occurred.
What test result follow-up requires after “seen”
Medicine often treats acknowledgment as a useful endpoint.
Message received.
Result viewed.
Alert opened.
But patient safety requires something more demanding.
In test result follow-up, the Agency for Healthcare Research and Quality describes “closing the loop” as a process in which a result is sent, received, acknowledged, acted upon, and communicated appropriately to the patient.. A result that has merely reached an inbox has completed only part of that journey. 4
That distinction seems obvious when written down. In practice, healthcare is filled with boundaries.
A specialist orders the test, but the finding belongs to another specialty. A covering physician sees the result while the usual clinician is away. A radiologist recommends follow-up but cannot arrange the entire pathway. A primary care doctor assumes the ordering specialist will act. The patient sees the report in a portal and reasonably assumes that if it matters, someone will call.
Every assumption can be understandable.
Together, they can leave an alert standing alone.
Studies of diagnostic delays associated with health information technology have found exactly these kinds of failures, including absent coverage for clinicians who were away and situations in which both a primary care physician and specialist were notified, yet neither took the necessary action because responsibility was ambiguous. 5
This is one expression of Medicine’s Blind Spot: technical capability can advance faster than the human system required to carry what follows.
The danger, then, is not always that information has been lost.
Sometimes the information is everywhere.
What is missing is ownership.
Ownership is a design decision
For years, I thought about responsibility mainly as a professional virtue. Good clinicians follow results. Good teams communicate. Good doctors do not let important findings disappear.
I still believe that.
But experience has made me less satisfied with virtue as a safety system.
A well-designed system should not depend on every busy person remembering every loose end.
Ownership should be built into the path of the alert.
That does not mean every notification needs to land personally with a physician. Many do not. Different messages can appropriately be routed to nurses, pharmacists, administrative teams, specialists, or other trained members of a care system. One large integrated health system has experimented with classifying millions of patient messages and directing them toward the role best suited to respond, rather than allowing every message to accumulate in the physician’s inbox. 6
The principle matters more than the particular technology.
When an alert is generated, somebody, or some explicitly defined team, should know whether it requires interpretation, what timeframe is acceptable, what escalation route exists if action does not occur, and how the system will know that the loop has finally closed.
This is different from simply sending the alert to more people.
More recipients can increase awareness.
Only ownership creates accountability.
The distinction will become more important as artificial intelligence and continuous monitoring generate still more signals. A system that detects twice as many abnormalities without deciding who must respond may not have doubled care. It may simply have doubled unresolved responsibility.
Technology can help prioritize, route, suppress, summarize, and escalate. Those are valuable functions.
But an algorithm cannot solve an organizational question that the organization itself has refused to answer.
Who is responsible now?
What the patient should not have to carry
There is another person in this chain whose role deserves particular care.
The patient.
Patient portals have made medical information more accessible, which is an important achievement. Patients should not have to wait passively for their own data.
But access and responsibility are different things.
When my patient’s wife pointed to the unexpected finding on the report, she had information. What she did not have was the context to know whether the finding required action that afternoon, the following week, or several months later, nor which clinician was supposed to arrange it.
Making information visible to patients should increase agency.
It should not quietly turn them into the final safety net for failures of clinical ownership.
The patient should be able to ask, “What happens next?” without first having to determine which part of the healthcare system is supposed to answer.
That may be the simplest test of an alert system.
Not whether the notification was sent.
Not even whether someone opened it.
But whether, after the alert appeared, the patient’s care actually moved.
The scan in that composite story had done its job. The radiologist had done the important work of recognizing and reporting an unexpected finding. The electronic record had transported the information exactly as designed.
Yet the patient’s question remained remarkably simple:
“Was somebody going to call me?”
That may be the question every alert system should be able to answer before the patient ever has to ask it.
The safest alert is not the one most people can see. It is the one someone knows is theirs.
*The opening scene is a composite drawn from recurring patterns in outpatient care. Details have been altered to protect privacy.
References
- Singh H, Thomas EJ, Mani S, et al. Timely follow-up of abnormal diagnostic imaging test results in an outpatient setting: Are electronic medical records achieving their potential? Archives of Internal Medicine. 2009;169(17):1578-1586.
- Murphy DR, Meyer AND, Russo E, Sittig DF, Wei L, Singh H. The burden of inbox notifications in commercial electronic health records. JAMA Internal Medicine. 2016;176(4):559-560.
- Holmgren AJ, Apathy NC, Adler-Milstein J, Bates DW, Rotenstein LS. Trends in physician electronic health record time and message volume. JAMA Internal Medicine. Published online February 24, 2025.
- Agency for Healthcare Research and Quality. Diagnostic Safety: Closing the Loop. AHRQ diagnostic safety resources.
- Howe JL, Adams KT, Hettinger AZ, Ratwani RM. Electronic health record usability issues and potential contribution to patient harm. Related analyses of health information technology associated diagnostic delays and root cause findings. JAMA Network Open. 2020.
- Liu VX, et al. Content of patient electronic messages to physicians in a large integrated system. JAMA Network Open. 2024;7:e244867.
