In reference to comments about the graph on the blog “How to Tell the Productive Surgeons from the non-Productive”
I agree, it’s a terrible graphic, but it’s not the real graphic. The ‘real’ graphic is actually a set of data that can use any of the ‘milestone’ times as a baseline for all the other times involved in an individual case. By choosing the baseline, discrepancies in groups of sequences are more readily seen, and the effect of one part of an operation on prolonging another part to cause a net loss or gain in time until completion can be played with so as not to just assume that saving time on part of a procedure will cause an overall gain by the end of the procedure. It also permits playing ‘what-ifs’ with altering the sequence of actions within a procedure to get a consistent and reliable milestone that has enough time to queue a second procedure to begin its setup time. It’s much easier to show with the program I use than explain in words or by a single graph on a blog.
Now, for some rambling….
As to what SPC is, where is the line drawn between fishbone diagrams, 2-sigma lines, patterns of 3 above the line, etc. and that of hard core probability? It’s all about detecting patterns to see if there is a tendency that you need to adjust or to use to your advantage . Classic charts like those of Shewart are suggestions that were applicable to the processes that people were trying to control. When you get into healthcare, the interactions can be much more complicated, and the patterns may need to be more complicated to show as much information as possible at once, but that can be understood, with practice, by someone who can and will alter some action to effect a beneficial change. There are books on my shelves that are compendiums of specialized graphics used by different industries. Also, there is lots of software nowadays that tries to graphically represent information by use of our sense of color, shape, relative position and size, fluctuations and more. One good example is Tinderbox by Mark Bernstein (www.eastgate.com) that permits such control of visual cues that it has a dedicated following of people who use it for visual exploratory analysis of data. This then invites the question of when should one use a narrow SPC chart instead of other exploratory techniques which are consistent with the ideas of process discovery, improvement and control.
Healthcare is not yet a fine art, at least in the OR. Anesthesia has always been considered half art and half science. I’ve surgeon friends who say the same thing about surgery. If the half art, half science idea is true, then exploratory analysis is essential and any technique—numerical or graphical–that readily shows patterns (hopefully with better understanding) that leads to better outcomes should be pursued. The ability to rapidly explore different ideas to clarify complicated interactions seems most easily done with visual cues (often graphs, but not necessarily graphs). Edward Tufte’s work was just the beginning. Look at the development of software games; take the Wii for instance, or take the forced feedback joysticks, or the auditory cues for World of Warcraft. A lot of work is ongoing to maximize the presentation of data in a form that our human senses are adept at analyzing.
So, I’m saying that ‘process control’ is still a developing field and its techniques are being furthered not just by ‘process control experts’ but by numerous others who are developing tools for its use.
Then there’s the discussion of integrating healthcare process control with financial analysis to make healthcare more profitable and/or affordable. That get’s to be real fun. How’s your finance background?