Measure and Maintain
From: M. D. Anderson Department of Performance Improvement
Date: January 15, 2008

Wayne Fischer: Hi I'm Wayne Fischer, the Manager of Research in the Institute for Healthcare Excellence here at M. D. Anderson. Today we're gonna review measure and maintain. Some of you may recall I did this a year ago; these slides are on our intranet so you can get the content there. Measure and maintain is also known as hold the gains or sustain the gains.

A good place to start is with the Shewhart cycle. Plan, do, check, act, some call it the Demming cycle where you adopt, modify, or abandon a change that you've tried after you've taken some measurements, you discuss captured lessons learned and a lot of people think job done, change work, keep with it, great. Well no, not yet because you need a second cycle and you might think of it as the SDCA cycle, that's the standardized do, check, act.

This is important because what experience has shown is that too often once a change has been tried on a small scale and shown to work that people tend to slide back into their old way of doing things, the familiar way, the comfortable way of doing things and you've lost the gain over time before you know it. It's sort of like the second law of thermal dynamics, kind of simply stated is that " ... Any system left unattended will seek it's natural state of maximum disorder because it takes the least amount of energy to maintain", and this happens in human systems, work systems, combined systems, it just happens unless you expend purposeful energy to hold in the gains that you achieve at such expense using some kind of quality control method. What might that be?

Well make the new way the usual way, make it permanent, make it a habit, get everyone involved in that process so they're using the same procedures and these procedures should be documented not only with narrative and text but with flow charts, train people, your workers, to these new procedures and then measure, put in some metrics. Your documentation as I mentioned, could be in flow chart form or visual queues. The(n) train well. Chain your teams together. Just generate new ideas about how to hold these gains that you've worked so hard to test and verify and on the measure, OK.

What about these measurements? How do we do it? Where, who, takes the measurements? Let's talk about that, metrics. Think back over your plan, do check act cycle, what were the key indicators of your processes' health? Its vital signs so to speak, even before you start your create solutions journey, what were the symptoms that prompted the action? Maybe you should monitor those too under the new way of doing the work. We can talk about qualitative and quantitative indicators. On your flow chart look at where there are critical junctures, where there are handoffs, where there are decisions made? That's clues as to where to put metrics down to monitor how the new way of working is doing. And look for those handoffs, again those decision points and error checking points.

Set the control standard for each process measure. Set a target and it can be quantitative or it can be qualitative. If it's quantitative you typically want to give a numerical value with a range because everything varies, don't forget that. If it’s qualitative have some descriptive criteria. One way to look at this is what's called operational definitions and you can find out more about that by talking to me or Duke or anyone else.

Set the control standard for each person. It may be that different parts of the work is done by different people so set the control standard for them. Expectations should be defined and described, the performance should be known and measured and compared to that standard. Is the worker enabled? That is do they have the tools and the training to carry forward on that job. Are they empowered? Can they make the decisions they need to make in those tasks? Are they stimulated to do the work? And define corrective actions. If the performance is not up to those standards, if there's a gap then you need to define what the corrective actions will be. Audit those controls. In other words check those checks that you've built in now. Are they working? What are the results? Compare them to those targets. To whom are the results reported?

Because experience shows us that if a worker in a task is producing results and measuring them that no one's paying any attention to the measurements, then they're gonna drift away back into their old comfortable way of doing the work and document that QC method. This could be thought of as the attentiveness, the purposeful energy I talked about earlier, being expended, to hold the gains that you worked so hard to achieve.

Now everything I've covered up to this point is a review of what I presented last year and as I said it's available on the intranet. Duke challenged me to add some more to this and I gave it some thought, did some more research, what else and how else I call it.

Another way to try and hold the gains is to implement changes that will make the new way the easy way to do the work. In other words implement ways that will make the old way difficult to do the work. And test those ways once you implement them, see if they really do work. A method that the Japanese developed they call 'pokayoka' just means fool proofing. That means to put in physical or mental ways that there's only one way to do the work, so you can't veer off the path the work process flow path so to speak.

Another way is to put in visual queues if your work process has to do with the work space around you, maybe visual queues will help people follow the new way, and remove the old way visual queues.

Another method is called EMEA, some of you may have heard of FMEA, failure modes and effects analysis, EMEA is error mode and effects analysis. It's a contingency planning tool, you're basically now looking ahead and asking yourself what could go wrong, what errors might happen? This is a prospective rather than look backwards, you're looking forwards, what could go wrong and how could we prevent it from going wrong? And you might design and implement job aids, manuals, diagrams, flow charts, computer programs, for example.

A deployment flow chart gives you a view of not only the different tasks that are required in the new way but it also lists the different people who are responsible, or the work positions that are responsible for each of those tasks. And so now you know who's supposed to be doing what, when in that new way of doing things. One of the computer programs you might think about developing is called an expert system where you capture the knowledge of the workers in the process as well as the new way of doing things, you integrate that in one computer and it's somewhat like a decision tree but a little more sophisticated but it guides the worker through the process and shares the expertise of all those who are familiar with that new way of doing things.

Another approach is to institute positive rewards for conformance and negative consequences for non-compliance. Now use this sparingly, I say tread lightly with this because you're not looking to punish people you're looking to reinforce good behaviors following the new process but sometimes they refuse to follow the new way for whatever reason, may be you need to take them offline and ask them what's going on, what's behind it and see if you can get their standard up to where it needs to be on the performance.

Well that's all I have for today, let's take it back to the studio. Duke and Tina.