In an IHI seminar this week, we’re discussing ways to use patient experience data. Two seminar case examples look at the use of tablets in outpatient settings. The organizations seek to capture patient responses to a series of questions, hoping that every patient every day will answer the questions. The organizations each want their clinic managers to get daily information about patient experience, both to mitigate service failures and to monitor the impact of improvement projects.
The seminar discussion includes the Prius Principles I described in a recent post.
The Prius principles outline five conditions for managers to use data to improve a system.
People have to
(1) sense the data: see, hear, smell, touch, taste;
(2) interpret the data: what do the data mean?
(3) connect interpretation to timely actions that can improve performance;
(4) have the power to act based on (3): "be in the driver's seat"
(5) actually act.
These conditions seem necessary but are not sufficient. In addition to the five conditions, there has to be some data plumbing built and maintained to define and capture measurements that comprise the data sensed by managers.
For example, in a Toyota Prius, Toyota engineers designed the sensors and wires to capture and present data on mileage performance in a dashboard display and then the Toyota assembly plant put all the pieces together.
In the two cases in our seminar, the data plumbing is the tablet hardware, the data storage and software to make everything work. Tablet hardware and software gather individual responses and place them into storage; more software aggregates the patient responses and then produces reports and messages sent to managers or accessed on a web page.
Relative to their traditional patient experience surveys, the organizations seek to increase both the volume of patient responses and to drastically reduce the time delays between service event, patient response, and data summary.
The organizations have so far invested the majority of their time and attention on data plumbing and consequently have paid less attention to the other five Prius conditions.
Since I’m convinced of the merit of the five Prius conditions, I tried to understand how this behavior might make sense.
I found at least one theory that explains the focus on data plumbing as rational: If the data plumbing is the weakest link in a data feedback system for clinic managers—data are too sparse and too delayed—then Goldratt’s Theory of Constraints says to focus on that weakest link and make it stronger. That’s the behavior in the two cases.
Goldratt’s theory also makes a prediction: When the data plumbing is no longer the weakest link, another step or resource in the system will then become the weakest link.
In both seminar cases, it looks like the new data plumbing will soon deliver more information than the managers can use productively. Thus, the next weakest link may be how well clinic managers can interpret the patient experience data in order to take timely actions that improve patient experience. Goldratt’s advice? Strengthen the new weakest link.
That’s my advice for these organizations, too.
More on Goldratt’s Theory of Constraints
In the 1980’s, Eliyahu Goldratt codified a management approach, the Theory of Constraints, that focuses improvement activity on the weakest link in a production system. (See the Wikipedia article: https://en.wikipedia.org/wiki/Theory_of_constraints accessed 29 November 2015).
The system might deliver a product or service; Goldratt’s theory certainly applies to the feedback systems in our seminar cases.
Here’s Goldratt’s specific insight: To increase performance of the system, you have to strengthen the weakest link. If you work to improve links other than the weakest, you’ll spend time and effort but not get any better results.
Goldratt outlined a five-step “focusing process” for improvement:
(1) Identify the system’s constraints.
(2) Decide how to exploit the system’s constraints.
(3) Subordinate everything else to the above decisions.
(4) Elevate the system’s constraints.
(5) If in the previous steps a constraint has been broken, go back to step 1, but do not allow inertia to cause a system constraint.
Goldratt’s theory helps me understand why people can focus on different parts of feedback systems and report improvements—different systems will have different weakest links.
If you are still challenged by reducing visual clutter in your organization, Toyota has just raised the bar.
Last Tuesday’s New York Times reported on the start of production of Lexus models at Toyota’s Georgetown, KY facility, the first time Lexus models have been manufactured outside of Japan.
Admired worldwide for the relentless focus of its production system, the article describes Toyota’s preparation to handle the fabrication of the Lexus model ES 350. The article describes investment in training and education, including sending workers to dealers in the U.S. and to Lexus factories in Japan—not surprising to anyone with even a modest understanding of Toyota’s approach to building cars.
But the new production line has another feature. It’s really quiet:
“The quest over the last 30 months has been to create what was unimaginable not too long ago: a largely noiseless, hushed atmosphere to house the new assembly line. ‘We want our team members to be able to hear a click,’ Mr. James [president of Toyota Motor Manufacturing, Kentucky] said. Toyotas have been produced in Georgetown since the first American-made Camry in 1988. But Lexus has a set of rules all its own. It is not just enough for a worker to see a potential problem; the worker should be able to hear it, too. ‘It’s about detecting things at a much deeper level as a vehicle heads down the line,” Mr. James said. ‘So we had to design the plant to allow for that.’"
As Mr. James points out, however, the aim is not to have a perfectly silent factory—just quiet enough to hear the appropriate click of parts assembled correctly or the different click of pieces that are not well matched.
What are the sounds in our organizations that demonstrate that work is going well? Can we hear them?
Back in 2009, I thought about “feedback leads to reduced energy use” and called it the Prius effect, in honor of an observation by Carrie Armel at Stanford, quoted in an article by Dan Charles (Science 14 August 2009: Vol. 325 no. 5942 pp. 804-811):
“The next big force for behavioral change may be technology that brings consumers face-to-face with their energy consumption. A simple version of such energy feedback is the dashboard of a Toyota Prius hybrid car, which displays the rate at which the car is burning gasoline. No one has carried out a controlled study of how drivers react to it, but ‘every person I know who has a Prius, they get a big grin when I mention feedback, and they have to tell me their personal story about how they’ve reduced their energy use,’ says Armel.”
I continue to see Prius effect thinking in range of projects—I fall into the same thinking myself: effective, attractive data served up to people is the key to driving better system performance.
No, it’s not.
Let me explain.
I don’t doubt that drivers of cars with energy information displays, like the Prius, can use the feedback to modify behavior. As a driver-owner of Honda Civic hybrid cars, I have used real-time mileage feedback to try to improve my energy performance while driving.
Here are a few reasons why the ca. 2009 Prius effect observed by Armel worked for people in her circle:
(1) The instantaneous mileage performance is available at all times with a glance at a gauge;
(2) The feedback signal in the early model Priuses was a bar graph that moved beside an indexed scale—higher is better, lower is worse, and it was easy to interpret.
(3) The main control lever in a car with automatic transmission is the accelerator; it’s easy to learn the connection between using the accelerator to reduce RPM and the mpg display.
(4) The Prius driver sits in the driver’s seat, interacting with accelerator and brake to change driving actions to improve performance.
(5) Prius drivers may be motivated by bragging rights and concern for the environment actually to drive conservatively, adjusting their driving actions, guided by their display.
Reasoning by analogy to the Prius effect, in the past 15 years people have designed many information interfaces to make energy use visible to building occupants and staff in order to drive action.
There are too many attractive displays, interactive dashboards, and web apps that fail to achieve real impact, which means changes to the system and improved performance. I've built my share.
And the tools to develop interesting interactive displays are increasingly powerful and flexible, making it ever easier to create more dashboards, web apps, and feedback systems. Nonetheless, achieving changed behavior and improved performance remains the aim of all that information design work.
A bit more thinking has helped to clarify the connection between information displays and people acting to improve performance.
The five reasons I suggested for the Prius effect are specific instances of a more general set of factors. Once someone has defined a measure relevant to assessing performance and figured out a way to generate data for the measure, now what?
To use relevant data to improve performance, a person or group must
(1) sense the data;
(2) interpret the data;
(3) connect the interpretation to one or more suitable actions that will improve performance;
(4) have the power or potential to act, based on step (3) (“can do”);
(5) actually act to change system performance (“actually do”).
While there’s a lot of fun and challenge for people like me to develop relevant measures and data and then to craft inspiring and engaging ways to achieve steps (1) and (2), that’s all preliminary to the change and action steps (3) through (5).
In order to build more useful and engaging data displays that will contribute to better system performance, I've concluded that I need to pay more attention to how people will take steps (3) through (5) so my information designs actually contribute to better systems.