Most medical devices have some form of audio alarm functionality to indicate a problem has occurred. Such seems an entirely reasonable and prudent thing to have… but what happens when you have 942 alarms per day in a single unit at National Childrens Medical Center in Washington DC? That’s an alarm going off every 92 seconds.
- We know that sleep is critical for well being1… but how is a patient to sleep, or get decent sleep with that level of alarm noise?
- We know that an alarm system that cry’s wolf all the time will in short order be an alarm that is either ignored at best, or vandalized at worst. Fortunately in medicine, folks don’t have to worry about someone taking a 12 gauge shotgun to the alarm (I saw a spec requiring a test along this line once)… but reflexively hitting the silence button without thinking is.
- We know from aviation, that if there are too many alarms going off concurrently, its pretty likely that the most mission critical one will be buried in the noise of the trivial ones. Ie, if there are too many alarm bells during an emergency landing, the probability of missing the landing gear horn goes up exponentially.
- In firmware engineering land, we have tons of information, but we selectively share information with the user to avoid overwhelming them. In the legal world, keeping this information hidden opens the door to liability, so the default is to alarm the user, even if its of very low importance.The other side of the coin, is that sooner or later, the legal world is going to see revenue in adverse event lawsuits due to information overload… and this will foster some serious thinking as to the balancing that needs to occur.
From a tech point of view, all of the above are easy to resolve… but it requires a system level approach will all of the stakeholders talking to one another. This is a recurrent problem for just about all large scale institutional problems. Its not the technology, its the people and politics behind it.
One of the ways device firms try to get around the people and politics issue is to provide for vast levels of user customization. Such provides for an economy of scale, but it does open the door up for liability somewhat. As a result, most firms will come up with very very conservative default settings, with the idea being a user who doesn’t configure things won’t miss an alarm function, no matter how trivial.
Consider some cardiac monitor data from Stanford University Medical Center. Over a 2 month period, they were experiencing 883 alarms per day. 43% of those alarms were non-critical, non-actional. 38% were for non-treatable events, only 19 of the alarms were code blue…2
The obvious solution to this is to be less conservative with the alarm limits… but this is tricky. Some have suggested, we need to create multi-disciplinary teams, capture a bunch of data, have a ton of meetings, draw up new specs, have a ton of meetings, and then have manfacturers develop new products, have a ton of meetings, and spend a bunch of money to solve this. Perhaps in the future, we could have the alarms automatically customize themselves off the patients trending data. In other words, its a time and money problem… but wait.
Frank Block says: We discussed at the AAMI Alarms Committee that this technology not only exists, but it has been incorporated in most ICU Monitors for the last decade. In other words, the clinicians who asked for this feature almost certainly have that feature today…but people don’t know the features of their own monitors! (And they don’t know how the alarms work, or how the alarms are supposed to work, etc.)2
Yep…. There is not much new under the sun. The same politics that leads to unwashed hands, and jokingly inadequate infection control is likewise a driver of beep mode. This is not to the moon type stuff to solve technology wise, but it certainly is people wise.
As far as how to fix this… visibility of the problem on all sides needs to be out in the open in a huge way. The system as it stands does not, nor will it tolerate such in part due to liability and/or license revocation concerns. No one wants to fess up about a patient dieing as they hit silence when they shouldn’t have. Likewise, no one wants to fess up to missing a critical alarm as they were jugging 10 others. Patients make lots of noise, but no one who can effect change is hearing. In a lot of ways, this was how aviation was run up until the mid 70s when ASRS came into being.
As aviation and medicine are worlds apart, I’m not if this would work… but a similar program which grants immunity to reporters and encourages sharing might be a good step forward to resolving this.