What is reliability? (Part 1 of 5)

Frank Federico, RPh, IHI Executive Director

Hi. I’m Frank Federico from the Institute for Healthcare Improvement (IHI). I’m an executive director here, and my focus is on patient safety and reliable design. Welcome to the first of a five-part series on IHI’s reliable design methodology.

What is reliability? What does it mean to you? Think of an example of a reliable process. You might think of one of the fast food chains, or you might think of Starbucks. My example is I visited Starbucks on the West Coast, I visited Starbucks on the East Coast, and then I visited Starbucks in Edinburgh, Scotland; and in each case, I received the same service, the same coffee, and even the pastries were set up in the same way.

There are other examples of reliable design as well. We might think of the computer industry and how reliable computers have become. We might think of how reliable our smartphones have become. We might look outside to other industries, other than health care, like the nuclear submarine industry, which are very highly reliable. Air craft carriers: very highly reliable. We might also look at hotel services. Many of those are considered very highly reliable.

Now, reliability only comes when it’s designed deliberately that way. That is, all of the processes have a context into which they are applied, and everybody does it the same way. You might consider them standard operating procedures.

So what might make these processes reliable? First, there’s a deliberate design. When the processes are designed, someone and a team get together and think through, “What are all the steps that are necessary, and how do we develop standard operating procedures that get us to the end point of providing great service to our customers, to our patients?” The entire team focuses is on that reliability, and when things don’t work well, they immediately work to improve on those processes. There’s constant testing, testing, testing. They continually refine the processes until they get to the point where the processes are capable of getting them the result they want. There’s also measurement. There’s constant, constant review of, “How well is it working? Are we getting the results we expected?”

Why is it that some processes are not reliable? There’s individual autonomy — that is, when individuals who are involved in the process choose to do it their own way. We tend to focus on physicians primarily, but this is true for anybody in health care or anybody else in any of the processes that we are trying to focus on. When individuals choose not to use the standard operating procedures, they now become the variation in the process. In order to have reliable design, you have to reduce that variation.

Now, there are times when somebody can do something different, but it’s important for us to know why that’s different. So, even in the standardization — as we talk about developing standard operating procedures — one of the things we will always focus on is the patient because we don’t want to harm the patient by using something that is good for everybody else, but not necessarily for that patient.

We tend to focus on benchmark performance. For the longest time, people who were on ventilators were at a certain rate of incidence of getting ventilator-associated pneumonia. When IHI developed the bundles, we then determined over time that many hospitals were able to go many, many months without ventilator-associated pneumonia. So the new benchmark was now zero. It was no longer the accepted rate of ventilator-associated pneumonia.

The other aspect that we worry about when we think about reliable design is an organization might have great outcomes, yet their processes are totally unreliable. They may have great outcomes because somebody is being heroic, or they might just be lucky that things aren’t going bad. So our query always is, “If your outcomes are really that good, are your processes reliable to support those outcomes?” because if in fact those processes are not reliable, then your outcomes might not be true outcomes.

We also, when we learn something is not going well, spend a lot of time on training, vigilance, and hard work. Now we know training is important. We know that people are vigilant, but after a while — human factors tells you that people can only perform at a certain level and then after that, their performance deteriorates. And we ask people to work harder. I think our health care people are working hard enough — asking them to work even harder is an impossibility. So although training is important, by itself training, vigilance, and hard work will not result in reliable processes.

Ultimately, many will say, “Well, we have a policy, and why aren’t people following the policy?” Policies are important. Policies are what an organization stands for. It’s the procedures that are necessary. Those are the things that have to be reliable in order to get the outcome that you want.

What is the IHI reliable design methodology? Well, the first thing is we ask you to start small. Use a subset of the population or a segment. We’ve tried to solve the problems with trying to care for all patients and then realized that one process does not work the same way for all patients.

We ask you to visualize the steps. Develop a high-level flow diagram so that you can see where the defects are and what you need to do to correct them. Identify those defects — and they may be existing in every box — but you have to identify which is the one that’s critical for us to begin developing the reliable process.

Think of a change concept, but think of things like simplification and standardization. Those are the key elements of improving reliability in any work that you do.

You also have to develop a backup plan because even in the best designed systems, sometimes the original process, the original standard operating procedure, will fail and will fail for some very good reasons. Having a backup plan ensures that the patient still gets the care that you expect the patient to get.

Also again, you have to test, test, test. All of the processes require that you engage people at the front line, test the particular process that you thought of, and then refine it with what you’ve learned from that testing, which, again, is all accompanied by measurement, measurement, measurement.

One thing to note about our design methodology — this is not intended for catastrophic processes. By catastrophic processes, I mean: When you are about to start surgery on the patient, you have to be right 100 percent of the time. You have to be able to start surgery on the right side. You can’t start on the left shoulder if you’re supposed to start on the right shoulder. If you’re about to administer blood, the reaction, the blood reaction when you have an incompatible blood type is significant and can cause death for the patient — so you have to be right 100 percent of the time when you’re about to administer blood. Same with chemotherapy.

The reliable design methodology works for non-catastrophic processes. These are processes where even if you fail, the patient is not going to be harmed in the next three to four hours or even the next day. You have a recovery plan that would enable the patient to then receive the care that you want and the patient won’t be harmed.