Reporting finds ‘alarm fatigue’ to blame in patient deaths

By Lizbeth Kowalczyk

In January, a patient at one of the country's best hospitals – Massachusetts General – suffered a fatal heart attack while staff did not respond to repeated alarms from his cardiac monitor.

State and federal health investigators in part blamed a phenomenon that is little known outside hospitals: alarm fatigue. In this case, 10 nurses on duty that morning did not recall hearing the beeps warning of the patient's falling heart rate. They told investigators they were desensitized to alarms after hearing them constantly throughout the workday.

Initially, I planned to just follow the Massachusetts General story as it unfolded. But after talking to several medical technology experts, it became clear that alarm fatigue is a growing problem without obvious solutions. And, it had not been written about much in trade journals, let alone the general press, meaning stories could have real impact.

My editors and I decided to try to figure out how often alarm fatigue is a factor in patient deaths, on a national level and in Massachusetts. The sources and techniques I used to report the stories – they were published as a two-day series – could be used by other reporters to write about alarm fatigue in their states, but also may be applicable to other topics.

We pulled the national numbers from the FDA's database of adverse events involving medical devices, called MAUDE. Reporters can search this database themselves – and may choose to do so if their paper has robust computer-assisted reporting staff and resources. But I decided to have ECRI, a nonprofit health technology research group in Pennsylvania, do it for me – searching for "death," "alarm" and "monitor" – because they already had developed software to search MAUDE. I didn't want to miss anything. In return, I agreed to credit ECRI in the story. They also are a great resource in general on medical devices.

A word of warning if you have not used the MAUDE database before: The reports are heavily redacted, incomplete and often don't reach a conclusion. They don't, for example, include names of patients, staff, hospitals or even states where incidents occurred. They do include incident dates and often colorful narratives written by manufacturers.  I found great examples for my story; instances in which alarms sounded for hours and no one responded, or where nurses silenced alarms – perhaps believing they were false alarms – without providing care to the patient.

We found 216 deaths over a recent five-year period related to alarm problems, many specifically to alarm fatigue, but most experts believe that number is vastly under-reported. I read each report and divided them into categories, such as "alarms sounded but were not heard or ignored by staff" and "monitor improperly programmed or turned off."

Those numbers provided the spine of the story, proof that the Massachusetts General death was by no measure isolated.

I used data from the state of Massachusetts to find actual cases that I could describe in my stories. Every state keeps data differently, but all states investigate hospital reports of untimely patient deaths and complaints. I requested all reports for 10 years from the Massachusetts Department of Public Health that involved a patient death and the following criteria: the words "alarm," "monitor," or "telemetry" mentioned in the report.  

The state came back with several dozen reports, both redacted complaint summaries and un-redacted "statements of deficiencies" written after on-site investigations. Not all of them were relevant, but most were.

 The state does some investigations for the Centers for Medicaid and Medicare Services and, in those cases, the reports can be released only by CMS. I filed a Freedom of Information request for these reports with CMS, but given how slowly the central office in Baltimore works, I was concerned that I would not get the reports in time for my story. So I reached a compromise with the Boston CMS office. The press person got permission from Baltimore to read me the reports, leaving out names. I urge reporters to request this assistance while waiting for written records.

I still needed patients and families. This was one of the hardest parts of the reporting, but ultimately the most satisfying.

I used dates of death and hospital names from the state reports to search obituaries and death notices online. I also had help searching death records at the state vital records bureau for anyone who died on those dates in those hospitals. I found a half dozen patients this way. I searched for relatives on Accurint. Ultimately, I interviewed the families of two patients who died because of alarm fatigue – I'd found both by doing this reverse search.

I also spent many hours in hospitals learning about monitors and watching nurses respond to alarms. I wrote a sidebar on what it's like for nurses to be besieged with alarms, many of them false. But in the end, this is an area of the story were I wish I had more immediate success. It was difficult to find nurses who would talk on the record about ignoring alarms, or nurses involved in an alarm fatigue-related mishap. Hospitals are very protective of staff, and rarely report these deaths to the nursing board – at least in Massachusetts.

One lesson I learned in reporting on this topic is to have confidence in what I know – even if hospital executives and patient safety directors seem to know little about the topic. Even if hospitals tell you alarm fatigue was not a factor in a death, if an alarm was sounding, and no one responded, chances are the hospital has a problem with desensitized nurses.

---

Lizbeth Kowalczyk is a medical and health care writer at The Boston Globe.