When Telemetry Creates More Noise Than Signal

Telemetry is one of those technologies that is so embedded in hospital care that we rarely stop to question the workflow around it.

A patient is placed on telemetry. The monitor continuously watches the patient's cardiac activity. Something crosses a threshold. An alarm fires.

Then what? Someone still has to decide whether that signal matters. And that is where I think the conversation around telemetry gets interesting.

Why Does Telemetry Generate So Many Alarms?

Telemetry is designed to detect changes. That is important. Continuous ECG monitoring can help clinicians recognize sudden cardiac arrest, identify deteriorating conditions, manage arrhythmias, and investigate symptoms such as syncope and palpitations.

The American Heart Association has established practice standards around which patients should receive ECG monitoring, how long they should be monitored, and how hospitals should manage the alarms those systems generate.

But sensitivity creates its own challenge. An alarm can be triggered by something clinically meaningful. It can also result from an electrode problem, a mispositioned sensor, an alarm threshold that is too sensitive, or another condition that does not ultimately require intervention.

Think about the workflow that creates.

• Every alarm is asking for attention.
• Every alarm creates a question.
• And someone has to answer it.

What Is Alarm Fatigue?

Alarm fatigue occurs when clinicians are exposed to such a high volume of alarms that they can become desensitized to them, increasing the risk that an important signal is overlooked or receives a delayed response.

This is not a new problem. The Joint Commission has identified alarm fatigue as an important patient safety concern for years. A study of non-ICU telemetry cited by JAMA noted that more than 90% of alarms in earlier studies lacked immediate clinical relevance.

The issue is not that clinicians stop caring. It's human factors. If a system repeatedly asks for your attention when nothing clinically meaningful is happening, you begin to build an expectation around the signal.

Most of us experience a version of this every day. If every notification on your phone were marked urgent, eventually the word "urgent" would stop meaning very much. In health care, the stakes are obviously much higher.

Does More Patient Data Lead to Better Monitoring?

Not necessarily. I think health care has spent the past several decades getting very good at collecting data.

Telemetry is one example. Bedside monitors, pulse oximetry, blood pressure monitoring, electronic health records, laboratory systems, and connected devices can produce a tremendous amount of information about a patient.

But data and understanding are different things. A telemetry alarm tells me something happened in a signal. What I really want to know is: what is happening with the patient?

The waveform gives me one piece of that story. The patient gives me another. I want both.

What If Telemetry Could Be Paired With Clinical Presence?

This is where I think the next evolution of monitoring becomes interesting. Imagine a telemetry alert occurs and the clinical team has the ability to quickly put eyes on the patient. Now we have another layer of context.

A clinician can see whether the patient appears comfortable or distressed. A virtual nurse or monitoring team can communicate with the patient or bedside team. The information coming from the monitor can be considered alongside what is actually happening in the room.

At Vitalchat, this is one of the areas we're interested in exploring: how do we bring real-time audio and video together with the data hospitals already have so clinicians can move more efficiently from signal to context to action?

Can Visual Context Reduce Alarm Fatigue?

Visual context does not eliminate false alarms, and I would be cautious about suggesting that any single technology solves alarm fatigue. But it can change the workflow around an alert.

Right now, many monitoring workflows effectively say: something happened, someone needs to investigate. That investigation may require calling a nurse, walking into a room, finding another clinician, or escalating the alarm before anyone has enough information to understand what is actually occurring.

What if some of that context were immediately available? The opportunity is to make the next clinical step easier.

This fits into a larger idea that I think health systems need to embrace: we should stop treating every technology as its own destination.

The real value comes from bringing these systems together around the clinical workflow.

Why Does Telemetry Need a Human in the Loop?

Technology is very good at identifying changes in data. Increasingly sophisticated monitoring systems may also help distinguish clinically significant events from noise. The American Heart Association has noted that AI applications for in-hospital monitoring have the potential to reduce false alarms and detect clinical deterioration.

That's exciting. But I don't think the future of monitoring is a machine making every decision. I think it's technology doing a better job of getting the right information in front of the right human at the right moment.

• Sometimes that information is a waveform.
• Sometimes it's a lab result.
• Sometimes it's the patient's history.
• And sometimes it's simply being able to look at the patient.

The question should be how we bring those pieces together so the clinician can make a faster, better-informed decision.

From Continuous Monitoring to Continuous Understanding

We have become very good at monitoring patients. Now we need to get better at understanding what all that monitoring is telling us.

Telemetry will continue to be an important part of hospital care. The answer is not fewer insights into what is happening with the patient. The answer is a better signal. Better context. Better integration.