Looking forward to adverse events?
Date: January 20, 2011
Q&A with Stephanie Terezakis on the use of failure mode and effect analysis to prevent errors in radiation oncology.
In their quest for patient safety, health care organizations need not only to learn from adverse events that have occurred, but also to anticipate and prevent errors before they happen. The Department of Radiation Oncology at The Johns Hopkins Hospital has taken a structured approach—known as failure mode and effect analysis (FMEA)—to tackle potential errors, such as treating a patient with the wrong radiation plan. Stephanie Terezakis, an assistant professor in the department, explains this tool and its limitations.
Why did the department decide that it needed to focus on errors that hadn’t occurred?
Our process is extremely complex, containing nearly 300 steps between the time that the patient arrives for consultation to treatment. Between those points there are multiple sub-processes, such as simulating treatment or contouring tumors. If you sit and wait for an error to occur at just any one of the hundreds of steps, and then try to prevent that same error from recurring, it’s unrealistic to expect that you’ll be effective. The next error will likely occur at a different step or in a different way.
We turned to failure mode and effect analysis to help us to systematically look at these potential errors.
How did you get started in this analysis?
We brought together a team with representatives from all of the different groups of providers who touch the process. We held long sessions of brainstorming about different types of potential errors—the “failure modes”—and came up with great insights that wouldn’t have occurred to a smaller, less diverse group of people.
We scored potential errors by severity, potential frequency and detectability. Those three scores are then combined into a composite number, known as an RPN [risk priority number], and ranked. You can then use those rankings to prioritize which weak spots to attack.
How did the team decide which fixes to use?
We also identified solutions to these high-risk failure modes and scored them according to feasibility and effectiveness. For example, one of the failure modes is that a radiation therapist might bring up the wrong treatment plan for the patient standing in front of them, and then deliver the wrong radiation to that person.
What was the solution?
We began requiring that the bar-coded, orange patient identification card—which all of our patients receive—be scanned at the radiation machine console before their treatment plan comes up.
What has your experience shown you?
You can’t try to fix every possible mistake. We’ve done two projects, one last year. Each showed about 130 possible ways for errors to occur. Our efforts focused on 15 different weak spots, and we selected 33 interventions for them.
Did you expect that this analysis would identify all potential errors?
No. After our second failure mode and effect analysis, we looked at our incident reporting system over a few months in late 2009 and found that roughly 40 percent were things that we had not considered in the analysis. For instance, there was a near-miss in which a patient had received radiation at another facility, but that prior treatment had not been properly communicated among our staff. As a result, our treatment plan at first did not take that earlier radiation into account. The mistake was caught before it affected the patient.
We’ve since begun the process of creating a checklist for each step in the workflow, and that checklist will include a field for recording whether there has been prior radiation.
What was your reaction to this finding?
At first we asked ourselves whether we had performed the process correctly. However, it wasn’t that surprising, considering the complexity of radiation oncology.
The result underscores the lesson that prospective risk assessments can complement, but not replace, reports of adverse events that actually happened. There’s a big push now in the field of radiation oncology to create a national reporting system, similar to the public database that the airline industry uses. My partner in the failure mode and effect analysis, Eric Ford, is heavily involved in a project to create such a reporting system. Hearing about events that happen at other hospitals can help us learn about our own systems and potential for error.