A CONVERSATION WITH DAVID CLASSEN, M.D.
This versatile physician holds the view that the Institute of Medicine actually understated the number of medical errors. He also doubts that the usual prescriptions for reducing errors will be effective.
David Classen is a clinical infectious disease physician, a medical epidemiologist, and a medical informaticist. His work in improving patient safety has included use of information systems to avoid adverse drug reactions and advance appropriate use of antibiotics in hospitalized patients. He brings those talents to his latest position, vice president in First Consulting Group's performance improvement group. Though now working with clients rather than at one health system, Classen says his mission remains the same: "Helping organizations build programs that measure and improve the quality and safety of the care they deliver."
Before joining FCG, Classen chaired the Institutional Review Board at Intermountain Health Care in Salt Lake City and Intermountain's Clinical Quality Review Committee for System Wide Drug Usage. He helped lead Intermountain in a shift from a department-oriented model to one built around more accountable multidisciplinary clinical business units. He received his B.A. and M.D. degrees from the University of Virginia, and a master's in medical informatics from the University of Utah, where he is an associate professor in the department of medicine. He spoke recently with Senior Contributing Editor Patrick Mullen.
MANAGED CARE: How accurate a view of the problem of medical errors and patient safety did the Institute of Medicine's report provide?
DAVID CLASSEN, M.D.: The IOM report, which is based on the Harvard Medical Practice Study, estimates anywhere from 750,000 to slightly over a million adverse events every year, and from 44,000 to 98,000 deaths resulting from those events. That particular Harvard study was extended to Utah and Colorado in 1992. We were able to compare its methodology to a computerized prospective screen for adverse effects. We found that the study underestimated the problem of adverse effects fivefold. So it's a much larger problem. The Institute of Medicine report woefully underestimates this problem.
MC: What are the largest types of adverse events?
CLASSEN: The largest categories are problems related to medications and procedures, and operative-related problems.
MC: Even if the IOM report understates the scope of the problem, what has been its effect on the thinking of health system executives?
CLASSEN: In the past there was a feeling among these executives that their organizations provide very high quality care, despite the occasional terrible tragedy. Mainly, the view was that the scope of the problem was not that great. The IOM report mainly picked up on work that had already been done. There's really not much new data there. Still, it tells health executives that they're living with a false sense of security, because their own reporting systems are not up to snuff. They're worse, far worse than even the Harvard Medical Practice reporting system, which is far worse than active ongoing prospective surveillance. Clearly, the IOM report challenges health executives to no longer believe their incident-based reporting systems.
MC: What's wrong with their systems?
CLASSEN: The problem with today's reporting is that it relies on an individual to detect a problem and then go through all the trouble to report it. Incident-based reports are incredibly laborious.
MC: So someone in a hospital who is swamped with work is being asked to take time out and fill out these reports?
CLASSEN: Yes, the health care system demands that everyone in it be more and more productive. On top of that, you are faced with filling out this report that takes forever and has to be signed by everyone on the planet. Why would you do it?
MC: You also may be drawing attention to a mistake you might have made.
CLASSEN: So why would anyone want to do this? They don't. We did a large study published in the Journal of the American Medical Association. We looked at incidents of adverse drug events in over 20,000 hospitalized patients by traditional incident-based reporting over an 18-month period, compared to a more aggressive computerized surveillance. We found that nine adverse drug events were reported during the 18-month period by traditional reporting systems. During the same period, the aggressive reporting systems we put in found 731. End of story.
MC: Do you have any sense of how close the 731 are to actual total incidence?
CLASSEN: The 731 figure probably underestimates the actual figure by about 50 percent.
MC: So part of the answer is a system that doesn't require people to step aside from their daily routine to report adverse incidents.
CLASSEN: The conventional wisdom of many of the new experts who have little experience in this area is, "Oh, it's all punitive. Nobody wants to report anything that could be used as punitive." That's wishful thinking. In reality, many health systems have learned that even when you take away the punitive nature of reporting, if you still force people to report incidents — given all their workloads, given how it's set up external to the care they deliver — it doesn't happen.
MC: How does a computer-based surveillance system work?
CLASSEN: Such a system would work automatically. That means when a drug is ordered electronically and the nurse gives it, it is documented electronically. If she gives it two hours late, then that will automatically be documented as a problem.
MC: Does this presuppose an electronic medical record?
CLASSEN: Not necessarily. I'm saying that's one example of how computerized surveillance might work. Another example is when a laboratory value on a patient suddenly becomes quite abnormal. That could stimulate a warning that there's a potential problem that someone could follow up on. You don't have to have a complete electronic medical record to do that. A sudden stop order for a drug could prompt a warning for someone to follow up on that. Maybe there was a problem with this drug. You're not creating an external reporting approach outside of the process of how drugs are used. That's why that particular approach does work. There's no doubt about it. The challenge to us is how to build monitoring of a process into the process itself. Many organizations have adopted the view, "We just have to enhance our voluntary reporting." This approach can increase reporting for a while with education, but then reporting inevitably trails right off as interest fades. The false assumption is that reporting declines because of improvements. When we started working on safety in 1986, there was very little interest in the area and only a small group of researchers actively looking at the problem. Now suddenly there are numerous experts in the area of safety with few new studies to match increased interest.
MC: Granted, I can see where you might look askance at people who proclaim expertise they don't have, but isn't it a good thing that more people are focusing on this area?
CLASSEN: It's terrific that more people are now focusing on this. What worries me is that health care organizations may get bad advice and not focus on key areas that need great attention. You've got a consistent refrain that says "You need to focus just on a culture of safety," or "You need to focus on creating a nonpunitive reporting system," or "You need to focus on cataloging every single potential error."
MC: You mentioned some things that can be done to reduce medication errors. What can be done to reduce procedural and operative errors?
CLASSEN: The operative errors and procedural errors are very complex. We can redesign how medications are delivered in ways that support clinical practice rather than dramatically change it. When we get into the operative and procedural areas, then we're getting much closer to the heart of clinical practice, and to the way clinicians perform the care they deliver. You have to change hearts and minds at that point. That's why many organizations started with medication first, because it is not as much of a culture transformation. The staying power of the safety era will be proved or disproved when we get into operative- and procedure-related problems.
MC: What are some common operative errors?
CLASSEN: We published a large study in the New England Journal of Medicine about operative use of antibiotics to prevent surgical wound infections. We're talking now about something as simple as making sure the antibiotic is given within two hours before the surgical incision. It's amazing how often that doesn't occur. That's one very simple intervention that can occur, that can make a big difference in the operative area.
MC: No one had to tell banks to go from paper to electronics, but in health care it seems like there is not the willingness to use new tools. What is going on?
CLASSEN: I think there is a willingness to use new tools. Once again, the conventional wisdom says physicians don't use computers because they're very resistant to using new tools. I'd like to challenge that. The problem is in the design of the systems that physicians are forced to use. They usually markedly increase physicians' time to accomplish a task, and don't clearly provide a value to the physician. That is an enormous problem. The physician-order entry programs that are currently on the market are incredibly difficult to use. They're designed by people who are nonclinicians for the most part. They take the task of writing a prescription from a 30-second task to a 5-minute task. The question is: "What's the benefit?" We very often find that the key things that would be of benefit to the patient and the physician are not there. These include explaining the formulary of the patient's insurance plan, checking for drug interactions, checking for cross allergies, and making dosing adjustments. They would all provide value to physicians and patients, but they are not even there, because the systems are poorly designed. As we get better order-entry systems, physicians will have handheld devices that they can carry throughout the inpatient and outpatient settings that will contain that information. That would be helpful to physicians and very helpful to patients. Then you wouldn't have to pass the law.
MC: How can medical education be changed to produce doctors who are more inclined to think along the lines that you described?
CLASSEN: Good question. I don't know the answer. We could encourage training courses for our students and residents in what quality and patient health care mean, how it will be measured, and how they will probably be held accountable when they enter a practice. There's no doubt in my mind that the average practicing physician in 10 years will have the outcomes of his care measured. I firmly believe that if he doesn't meet a certain standard, the physician will not get reimbursed for the care he gives. Exactly how to do that in medical schools or residency training settings, I'm not sure.
MC: What can health care learn from industries that have to be highly concerned with safety, such as the nuclear power industry or the airlines?
CLASSEN: As industries evolve in maturity, the initial focus is cost. Later, quality becomes a major focus. We've seen that in the automotive industry. As automakers started to flatten out the differences in cost, quality became a major differentiator. In health care, as costs begin to flatten out the differences between organizations, the differentiator will be quality of care. I might argue that health care can learn more from other manufacturing industries than from nuclear power. I know it's very trendy to say that we should learn from the nuclear power industry and the airline industry. I don't think they have nearly as much in common with patient care as manufacturing industries. Manufacturers take a particular process and design it ahead of time for safety and quality, and build monitoring directly into it. Nuclear power and airlines don't do that.
MC: What current trends are candidates for the scrap heap and what trends are keepers?
CLASSEN: I think the idea that hospitals will disappear is going to the scrap heap. In the area we're talking about, I have no doubt that accountability for quality and outcomes of care will have a profound impact. Helping organizations to measure and improve the quality of care they deliver is a several-decade-long undertaking. If you look at other industries, they never stop doing it.
MC: How large a role should government play in patient safety?
CLASSEN: Well, it already controls 50 percent of the market, and it already has a major role on the regulatory side. The government will play a role in fostering research and education. It will play a role in setting up quality standards. I don't have any doubt about that.
MC: Why did you join FCG?
CLASSEN: I joined FCG because they wanted to build a practice that would do the same things we were doing at Intermountain. We are helping organizations measure and improve the quality and safety of the care they deliver, using information technology.
MC: Thank you.