Donald Berwick and FMEA

Driving in my car, I happened to hear Donald Berwick interviewed on NPR. Dr. Berwick is well known for quality improvement in healthcare – my previous and somewhat heretical observations are recounted and amplified here.

In, 2003, Dr. Berwick had an editorial in the Washington Post (July 29, 2003) and New England Journal of Medicine (2003;348:2570-2572) which I disagreed with in part, because it was too dismissive of root cause analysis. My response was: There is nothing wrong with the concept of a root cause. Int J Qual Health Care 2004;16:263.

I also objected that Dr. Berwick referred to one of his coworkers as “Tom Nolan, one of the leading quality-improvement scholars ofour time.” Now as for leading quality improvement scholars, I’ve heard of Deming, Juran, Crosby, and Taguchi but until mentioned by Dr. Berwick, I hadn’t heard of Tom Nolan. Scholars should have a body of work. One can view Nolan’s. It seemed to me that Nolan had been anointed his status by Dr. Berwick.

I also disagreed with how FMEA (Failure Mode Effects Analysis) was (and is) conducted at the IHI (Institute for Healthcare Improvement). Besides the classification columns, IHI has three columns: failure mode, causes, and effects. These columns are nested within “steps.” But these steps are not individual process steps. In the IHI example case of chemotherapy medication, the step “administration” implies many separate process steps but none are spelled out. A real FMEA starts with a process map which leads to a series of individual process steps. The process steps are the starting point of a FMEA and each process step can fail in different ways. Process steps can be procedure steps (process blood), detection steps (check for hemolysis), and recovery steps (request new sample). Classification should be only for severity and probability of occurrence but IHI also includes detection. What does detection of a detection process step mean? I don’t understand why people are trying to oversimplify a traditional FMEA (occurs in CLSI EP23 too). This oversimplification makes things more complicated.

In addition, there is the following problem. In a traditional FMEA, classification scores are obtained by multiplying severity and probability of occurrence and the scores are totaled across all events. IHI wants to see the total score reduced – this is the stated aim of the IHI example. Now the only way to do this is to lower probability of occurrence. While this sounds reasonable, most severe events already have probability at the lowest level, meaning that focus in IHI FMEAs will be on less severe events. This is somewhat clouded when you throw in detection as a classification but it is nevertheless a serious problem.

So my extremely limited experience with the execution of quality improvements efforts championed by Berwick is not good, but he is a charismatic speaker.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: