What Causes Alert Fatigue, and How Do You Make Urgency Mean Something Again?

Alert fatigue is usually described as a volume problem. The measurement suggests it is closer to a signal design problem. In a study of 2,706 secure messages sent to hospitalists, 99% were nonurgent and 66% were non-actionable (Clawson et al., Journal of Hospital Medicine, 2026). When almost everything arrives with the same weight, each notification costs a clinician a decision, and the truly urgent one has nothing to distinguish it.

Healthcare teams have more ways to stay connected than ever. Secure messaging makes everyday collaboration easier. Paging provides a path for time-sensitive needs. EHR notifications surface information inside the clinical workflow. Phone calls still matter when a conversation has to happen now. Each channel serves a purpose.

The opportunity is making sure the way we communicate reflects what a situation actually requires. A routine update, a question that can wait until later in the shift, and a time-sensitive clinical concern are all important. They should not ask for attention in the same way. When everything arrives feeling urgent, urgency stops carrying information.

What causes alert fatigue?

Volume is part of it. The larger part is that most notifications arrive without telling the clinician what they are supposed to do about them.

Every alert creates a small decision. Does this require action? Is it mine? Does it need me now, or can it wait? Multiply those decisions across secure messages, EHR notifications, pages, calls, alarms, and reminders across a twelve-hour shift, and communication becomes its own category of cognitive work.

The message data supports that framing. In a study of 2,706 secure messages across 365 conversations involving hospitalists, 99% were nonurgent and 66% were non-actionable (Clawson et al., Journal of Hospital Medicine, 2026). A separate study of 61,057 messages at another academic center found interns exchanged a mean of 39.8 messages per day with nurses, and that the median time spent reading a message was 35 seconds for interns and 26 seconds for nurses (JMIR Medical Informatics, 2025). Both are single-institution studies, so read them as texture rather than national rates.

Nothing in those numbers suggests the messages were unnecessary. They suggest that a channel carrying almost entirely nonurgent traffic is a poor place to put something that cannot wait.

How many clinical alarms does a patient actually generate?

Enough that the count stops being the useful question, which is why it is worth looking at what kind of alarms they are.

A cross-sectional study across one intermediate care unit and three intensive care units at a US academic medical center, covering 17,442 patient encounters, measured 9.36 alarms per patient-hour in the intermediate care unit and between 30.76 and 40.94 alarms per patient-hour in the intensive care units. In the same data, 88% of all alarms were technical rather than related to a change in patient condition, and 74.4% of all alarms were silent (Scientific Reports, 2026).

One caution on the widely quoted claim that 85% to 99% of alarms do not require intervention. That figure traces to an AAMI publication from 2011, which makes it more than a decade old, and definitions of "false" and "technical" differ across studies. The 2026 measurements above are the more current basis for the same point.

Do clinicians delay responding to alarms?

By their own account, frequently, and usually because they are already doing something that cannot be put down.

In a cross-sectional study of 3,986 registered nurses across 213 acute care hospitals in New York and Illinois, 83% reported feeling overwhelmed by alarms, 76% reported delaying response to an alarm because they could not step away from another patient or task, and 55% reported a situation where a patient needed urgent attention and no one responded to an alarm (BMJ Open Quality, 2023). These are self-reported perceptions from two states, collected in 2021, rather than audited event data.

That 76% is the operationally interesting number. It describes a clinician who heard the signal, judged it against what was in front of them, and made a reasonable call. No amount of louder alerting improves that moment. What helps is a signal that carries its own priority, and a path that continues on its own when the first person cannot take it.

"If there's too much alerting, or if it's not done in the right way, we can end up ignoring truly significant items," said Allison Weathers, MD, Associate Chief Medical Information Officer at Cleveland Clinic, describing an enterprise that was generating more than three million clinical decision support alerts a month before it formed a governance workgroup.

Does changing the signal actually change response?

There is measured evidence that it does, and it is one of the more encouraging findings in this area.

A retrospective pre-post study at four hospitals found that pairing perceptually distinct alarm tones with voice notification produced nurse response to code blue events that was 15% to 59% faster, alongside a 20% to 38% reduction in alarm burden (Hansen et al., Human Factors, 2023). Faster response and fewer alarms moved in the same direction, which is the opposite of what a pure volume framing would predict.

"We devised a way to save lives by getting nurses to the bedside of patients in cardiac arrest faster while eliminating the number of alarms nurses get," said Emily Patterson, PhD, professor in the School of Health and Rehabilitation Sciences at The Ohio State University.

What does the Joint Commission require for clinical alarms in 2026?

The requirement still exists, and its name changed, which matters if your policies cite the old one.

As of January 1, 2026, the Joint Commission's National Performance Goals replaced National Patient Safety Goals as part of Accreditation 360, and the alarm requirement previously carried as NPSG.06.01.01 now appears in the hospital program as NPG.01.05.01. The standard reads: "The hospital improves the safety of clinical alarm systems." Hospitals must identify the most important alarm signals to manage, considering risk to the patient if a signal is not attended to and whether a signal contributes to alarm fatigue, and must maintain documented policies covering settings, who may change them, monitoring and response, and checking alarms for accuracy (The Joint Commission, 2026). Any policy still citing NPSG.06.01.01 as current is out of date.

What does urgent communication need that a notification does not provide?

Four things: an owner, an expected response time, an acknowledgment, and a defined next move if it does not come.

Sending information and getting a timely response are different pieces of work. When something is time-sensitive, the workflow needs to answer more than where to send it. Who is responsible right now? How quickly is a response expected? How does the sender know somebody has taken ownership? What happens if they have not?

Consider a nurse who identifies a time-sensitive change in a patient's condition. The organization has already defined how that level of communication is handled. The request routes to the clinician responsible for that role at that moment, without asking the nurse to work out who is covering. The clinician acknowledges it, which gives the nurse visibility that somebody owns it. If acknowledgment does not happen inside the expected window, the workflow escalates according to the organization's process. Meanwhile routine communication keeps flowing through the channels built for it.

Backline helps organizations attach that structure to the communication they already send, so a time-sensitive request carries an owner, a response expectation, and an escalation path rather than depending on someone noticing it.

Structure here does not mean making more communication urgent. It means making urgency legible. A message may only need to share information. A task may need an owner and a due time. A time-sensitive request may need acknowledgment. An urgent need may require escalation until somebody responds. When those distinctions live in the workflow, clinicians are not deciding from scratch how to move each item forward. Because the routing reflects how your teams actually cover work, it adapts to your organization rather than imposing a template. Your workflows, your way.

The so-what

Clinician attention is one of the few genuinely scarce resources in a hospital. Communication technology should help direct it. Routine communication should stay routine, time-sensitive needs should carry clear response expectations, and when something truly is urgent there should be a reliable path to somebody who can act, including a next step when the first person cannot.

The goal is not fewer alerts for their own sake. It is clearer signals and escalation that happens when attention is actually needed. If you are thinking about how routing and escalation could work more intentionally across your organization, explore how Backline Pathways supports structured clinical communication that adapts to your teams and your priorities.

When everything is urgent, nothing is urgent. The opportunity is to make urgency meaningful again.

Backline Pathways connects routing, acknowledgment, and escalation to the work behind a message, so urgency has a defined path to response rather than competing for attention with everything else. If you’re thinking about how that could work across your organization, a Workflow Assessment is a good place to start.