CPOE failure modes and effects analysis brings up some good questions

A Failure modes and effects analysis (FMEA) is basically a methodology for predicting potential pitfalls in a project and preemptively finding solutions. This is in contrast to a root cause analysis (RCA) in which case you figure out what went wrong after the fact. Kind of like asking “what could make a plane crash and how to prevent it?” (=FMEA) versus “what made the plane crash and how do we prevent it from happening again?” (=RCA).

My current position is the first in which I’ve been involved in an FMEA, and I’ve personally found them to be powerful tools. We did an FMEA prior to implementation of our BCMA system and came up with what I thought was a pretty good list of things to look out for. Of course what the administration chooses to do with that information is a different story, but at least it’s available if needed.

I’m now involved in an FMEA for our CPOE project. It’s interesting that the issues associated with CPOE are much different than those associated with BCMA and much scarier. The scale of the CPOE project is several orders of magnitude bigger than the BCMA project. One of the biggest issues brought up during discussions on CPOE has centered on what to do when the physician selects the wrong patient or the wrong medication; notice I used “when” and not “if” because it’s going to happen. In the paper world an incorrect order is often found because the patient’s information on the physician order sheet doesn’t match the name on the chart. In the electronic world there’s no such safety. I also believe it’s significantly more difficult for the physician to grab the wrong chart than it is to select the wrong patient. I could be wrong, but I don’t think so.

How do you prevent a physician from choosing the wrong patient? Good question. Feel free to chime in at any point with a solution. One thing that crossed our minds was the use of some type of proximity system where the physician could only enter orders on a patient after scanning their wrist band or by using some type of RFID tag that automatically brings up the patient medical record when entering the room. I think it’s a valid idea, but implementation would be a bear. Throw in the fact that physicians don’t believe they can make a mistake and the battle is on.

Now assume the physician gets the patient right, but chooses the wrong drug. After all, when you search for captopril you’ll get more than one option if your hospital formulary contains more than one dosage form, tablet strength, etc. For example, depending on your system setup searching for “pro” could result in “Protonix” or “propranolol”. Both come in 40mg strengths, but they aren’t even close in what they’re used for; I’ve actually seen this error. Unless the medication ordered by the physician is wildly out of line you might never know what the intent was. The likelihood of something like this causing harm is low, but it could cause delays in therapy or possibly expose the patient to unnecessary medications. So how do you prevent this from happening? I personally think floor based pharmacists doing real-time medication evaluation and disease state management is the best approach here. Being involved with the patient’s care plan is the only real safeguard against something like this.

These are complex issues and I don’t have great solutions. If you have any ideas or have seen something in practice that might work I’d love to hear from you.

14 thoughts on “CPOE failure modes and effects analysis brings up some good questions”

  1. one solution would be to force indication fields for all med orders.
    would help id errors and all staff involved are supposed to know why patient is on med anyway and this info isnt always readily available

  2. Jerry

    You’re right that these are difficult processes to make risk-free. The best we can hope for is to find ways of minimising the risks. Besides scanning a patient’s wristband before being able to enter orders, what about having a photograph of the patient appear on the order-entry screen? In most cases the clinician will have seen the patient before ordering medication for them.

    As for minimising the risk of picklist selection errors, surely requiring the clinician to specify the indication for the medicine about to be chosen would help. Presumably, if the indication is specified before medication selection, you could get the system to include in the picklist only medicines with an appropriate indication (i.e. the picklist is constructed on-the-fly). That way, if a selection error is made there’s less chance of the chosen medicine being therapeutically inappropriate.

    Floor-based pharmacists is a nice idea and would certainly help to reduce the risks but I suspect few establishments will really manage to implement a system where such a pharmacist validates every medication order before it’s administered.

  3. @Colin – Thanks for the information. I appreciate the alternate solutions because quite frankly they’re better than what we came up with, and seem like a much more logical approach.

  4. Hi Jerry –

    Your discussion of FMEA interests me in part because I have been intensely focused on FMEA in healthcare for several years (I built a web-based FMEA system, in fact, as the result of my belief that FMEA is generally underutilized to improve clinical environments).

    Seeing your readers respond so intelligently to your post about the CPOE implementation gave me an idea: what if there were a kind of “open” FMEA in which your readers could collaborate, brainstorm etc on a general topic (such as CPOE implementation/processes) in a more structured way? Eg, you could lay out the process steps/substeps etc, and people interested in participating could contribute their thoughts on failure modes, causes, actions, etc… Capturing the ideas from multiple perspectives might produce some useful results.

    If you (or any of your readers) have any interest in trying this out, I’d be glad to offer up my web-based FMEA system to conduct the experiment – no charge, of course, as I think it would be really interesting to see if the idea could grow legs… My system could easily facilitate such a project. (description at http://www.qipath.com; I’m basically “the guy in the garage”).

    Just a thought – let me know…


  5. Hi Tom –

    I think you have a great idea. I think the entire FMEA process is underutilized as well. I’ll certainly ask around to see if anyone is interested. I have one colleague in particular that might find the idea worth pursuing.

    Thanks for stopping by and especially for offering up your web-based FMEA system. Just knowing that their is a web-based FMEA system out there is exciting. Before your commentary I had no idea.


  6. @Jerry Fahrni
    Jerry – FYI (and full disclosure), the Institute for Healthcare Improvement (IHI) also has a web-based FMEA tool that you can use for free – but I built mine because I found that tool lacking. People who have used both tools have told me they like mine better – so, of course, do I.

  7. Too funny. Thanks for the full disclosure, Tom. I would expect nothing less, that is about you linking yours better.

  8. Hi Jerry,
    How about creating a search engine for pt which eliminating the list of pt;s names as MD entering pt’s demographic information.ie:name, DOB, medical record number(MRN), room number….I think most errors occured because the pt’s field was populated automaticly with too few required demographic info. If we make MD CONCIOUSLY CHOOSE or MUST put in at least 3 pt’s ID info, then….

    About choosing the right med, REQUIRED INDICATION field is a MUST. I think. Unless MD want to use off label indication…that will be a problem.

  9. Hey Chau – I think both ideas hold promise. I really like the idea to require indication for each medication. Unless the off label indication is really weird I think most pharmacists would get the gist. Thanks for stopping by.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.