Advanced Search

  • Topic
  • Specialty
  • Content Type


Also Related

< Back To Patient Safety
0 dislikes

< Hide

Comments For

Patient Safety Data Driving Change: A Model Methodology


< Shrink

Add Your Voice

All comments are posted anonymously. Your comment will be attributed to: "Anonymous user."

post comment


Are you sure you want to delete this comment?

Patient Safety Data Driving Change: A Model Methodology

By Tom A. Augello, CRICO

Related to: Ambulatory, Clinical Guidelines, Emergency Medicine, Primary Care, Medication, Nursing, Obstetrics, Other Specialties, Surgery

Checklists, automation, simulation, drug abbreviations, or marking the surgical site—it all starts with error data and re-starts with outcome measures.

Guest Commentator

  • Robert Hanscom, JD; CRICO/RMF; Cambridge, MA
  • James Pichert, PhD; Vanderbilt University; Nashville, TN
  • Luke Sato, MD; CRICO/RMF; Cambridge, MA
  • Tom Snyder; Princeton Insurance; Princeton, NJ


Numbers are everywhere in health care safety. Data gathered through clinical quality measures…Malpractice claims… Patient complaints. But how can an organization use these patient safety numbers to reduce the human and financial costs associated with medical errors?

A gathering of medical malpractice insurers, risk managers and patient safety leaders met in Palo Alto, California, in late 2008 to consider the question. One of the answers was a model methodology described by two officers from CRICO/RMF, Harvard's patient safety and malpractice insurance company.

The methodology outlines six steps needed to drive change through a health care organization effectively—starting with numbers to identify specific risks, through measuring any change brought about by intervention.

Capturing the data is just a first step, and the steps don't always start and stop neatly in sequence. Regardless of the data source, the richer the coding scheme, the better able an organization will be to pull meaning and set priorities from the data.

Dr. Luke Sato is Chief Medical Officer for CRICO/RMF at Harvard. Dr. Sato used a case to illustrate how the data can be captured and coded so malpractice cases can be aggregated to identify trends and breakdowns in the process of care.

"I would like to walk you quickly through a case. We have a middle-aged gentleman who complains of chest pain and comes to the emergency department. And through this we have basically four types of categories, access, assessment, human factors and diagnosis. We basically have a set of questions that we ask consistently in this order to approach what coding and codes to use. So for this case, for example, did the clinical team, the team in the emergency department, misdiagnose. It could be a yes or a no, and with that there's certain outcomes and then I can show you the coding results. So here are some examples of the questions that we would use. So, for example, access, was there a delay….."

Getting an organization to take steps that will make care safer and reduce costs takes more than raw numbers, even dramatic raw numbers. Robert Hanscom is Vice President of Loss Prevention and Patient Safety at CRICO/RMF. Hanscom described how Step Two requires "framing," or setting a context for the data.

One concrete way to do this is with comparative data or benchmarking. And again, Hanscom used a case to illustrate how it can be done.

"And for this organization, as we are looking at their malpractice profile over the last five years, this is really what we were able to show them. You can see that their top categories starting from the left going over to the right are: general medicine first, medical subspecialty second, surgical subspecialty third, general surgery fourth, OB/GYN, orthopedics and emergency. So let me show you what we did with this organization. We have the benefit in CRICO, of course, comparing to CRICO peers, so that was the comparison of this organization to peer organizations within the CRICO system, so organizations that were similar to themselves, not community hospitals, other academic centers. So that was the first look. The second look was to compare them to academic centers from our much larger comparative bench-marking database, and you can see what that did. So now we're comparing them really in a much broader sense, not only regionally but across the country. And what that allowed this organization to do is to say all right, this is now giving us a lot more understanding, as far as where we are probably outliers in not a good way. What they did was, they actually said these are really our priorities here. These are the areas that we want to start with."

Other sources of data can help with the context or "framing step." Dr. James Pichert of Vanderbilt University School of Medicine related how Vanderbilt developed a database of coded patient complaints that complemented its malpractice data.

Dr. Pichert says that doctors pay much more attention to their own numbers and are motivated to make changes, when they see the results for their colleagues. Most physicians want to know how they compare to others.

"A physician can dismiss, as from a crazy person, any single complaint. But when you develop and show that there's a pattern over several years, and that I do stand out from the rest of my group, it can be very powerful. Sixty percent do better, just by being aware of where they stand out."

Complementary sources of data also help with the third step in the model methodology, which is called "Ask." In the "Ask" step, confirmation is sought to ensure that the problem still exists and whether it still looks like a priority. Bob Hanscom:

"We now get to step number 3, and with this organization, these were the very questions that we asked. The first one of course, are you still at risk? In today's environment, now in the past you were, let's talk about. So are you still at risk for unreliable receipt of critical test results? It had shown up in their malpractice data. It was the first question that we wanted to ask."

The answers can come through other data sources that are more contemporaneous, such as incident report data, patient complaint and resolution data, quality reporting measurements, or even practice evaluations. The fourth step in the model methodology is "Seek." Dr. Sato:

"'Seek,'" from our perspective means to look for interventions, research solutions, use the information that we have as well as best practices and knowledge that exist outside in the real world. What is actually working? And create an inventory and to apply that to the specific interventions."

Princeton Insurance Company has joined a growing number of similar malpractice insurance organizations in pooling their data for benchmarking, using standardized coding developed at Harvard. Tom Snyder is Vice President of Healthcare Services for Princeton Insurance. Snyder said that the deep analysis of its data creates a more specific question for the "Ask" step. Deep analysis also gives direction when it's time to seek solutions and go to the next step in the model, called "Act," to implement interventions.

"If we look at office systems, it's failure in follow-up, patient systems and that ties very well. So in other words, you send somebody for a colonoscopy and they never get it done and you don't know they never got it done because you don't follow it up. And when you combine that with the behavior related and that you see that it's noncompliance with the treatment regimen, I mean it fits very nicely. And so again, it's diagnosis related and it's cancer and it's getting the patient to do what they do and putting follow-up systems and for god's sake document what it is that you're doing. And if you just did it for those patients that you suspected that cancer might be an issue, it would be one really big step."

After an action is taken, the last, and perhaps most difficult, step in the Model Methodology is called "Measure." Bob Hanscom says start with existing metrics:

"What we try to understand is what data is the organization already collecting, that actually it doesn't add to their work. It doesn't add to anything other than it just helps them understand what piece of their data, what data set actually they should probably keep their eyes on a little bit closer and actually even have somebody responsible or accountable for monitoring whether or not the progress is really being made, whether or not the change has been sustained, whether or not that actually in the short term seeing improvement in that factor that was actually causing missed and delayed diagnosis in the malpractice cases…. We're not asking them to collect more, but we're asking them to look at your existing metrics and then pay attention to ones that are reflected in graphs and charts like these because these are the ones that are most responsible for the worst of the worst. The worst outcomes to the patients, these are responsible for the tragedies."

January 1, 2009
0 dislikes

< Back To Patient Safety