Quality Talk
Quality Talk
Brent C. James, MD, MStat, joins us to discuss the subject of clinical errors and patient injuries. You’ll learn why it’s important to distinguish between the two concepts. James is vice president for Medical Research and executive director for the Intermountain Health Care (IHC) Institute for Health Care Delivery Research in Salt Lake City. IHC is renowned as a national leader in quality care. Among James’ recent contributions to better patient care is his service on the Washington, DC-based Institute of Medicine’s investigation and subsequent report, "To Err Is Human: Building a Safer Health System."
Q. In the Intermountain Health Care sys-tem, how do you handle error prevention and reduction?
A. Patient injuries are a subset of general quality and clinical improvement programs. They correspond to a general series of clinical process improvements.
One of the most profitable things we have done is to think about them in process terms. In priority order they are:
1. adverse drug events (ADE);
2. hospital-acquired infection;
3. venous thromboembolism;
4. decubitus ulcers (bed sores);
5. the process of strength, agility, and cognition (compromised strength, agility, or cognition), which shows up as patient falls and injuries, and the use of chemical or physical restraints;
6. blood product transfusions;
7. appropriate management of our patients in extremis — basically, patients who are crashing;
8. a grab bag category of really unusual events that really are sentinel events: suicides, kidnapping from a newborn nursery, and wrong site surgery.
If you eliminate injuries in the top four groups on the list, you’ve probably eliminated 80% of your injuries.
Another important concept is to be really careful to say patient injuries instead of errors. The distinction between the two is really quite important. It effects a change in your thinking about these topics. We have been just as effective in addressing injuries that people would not classify as errors as we have been in removing injuries that people would classify as errors. So the term errors is not particularly useful.
To remove the injuries, you have to have an accurate way to detect them first. At LDS Hospital (the IHC facility in Salt Lake City), our senior medical informaticist, R. Scott Evans built three parallel systems to detect ADEs at the hospital:
1. The same system all big hospitals use to report to the Joint Commission.
2. "Enhanced reporting" designed by Evans to eliminate paperwork for the nurses. He reasoned that perhaps nurses were failing to report ADEs because of the extra work in filing the forms.
3. He programmed the electronic systems at LDS Hospital to look for any event or any clinical response that might be associated with an adverse drug event. For example, every use of an antidote drug, the most common being Maloxone to counteract an opiate. That was an automatic flag for a nurse researcher to review the case. Benadryl was another antidote that raised a flag, as well as sudden changes in lab values or pharmacy orders.
Evans ran those three systems for 18 months. Here are the results:
• Nurse incident reports revealed nine confirmed moderate or severe drug events.
• Enhanced reporting turned up 91. That’s 10 times more if you relieve the reporting burden.
• The computerized approach of health systems monitoring found 731. That’s 80 times more.
This is compatible with a large body of research showing that, when measured, nurse incident reporting grossly — massively — under detects. That means we don’t appreciate the size of the problem, the opportunity it presents, and we fail to respond as a hospital, as a system, or as a profession. So the first step is to accurately detect.
Q. What could institutions do to address the under detection of patient injuries even before more legislation comes on line?
A. Since we started the health systems monitoring approach, we’ve gone back to analyze the most common triggers that found moderate or severe ADEs. They don’t get mild ones, but those aren’t as critical to the patient, nor as expensive to treat.
If you manually track 14 parameters in your pharmacy, lab, and nursing notes (quite simply, diarrhea and rashes), then send someone to follow up on a daily basis, without any computer, you’ll pick up about 96% of all the ADEs I just mentioned. And for a surprisingly small expense. It’s far more efficient than chart review systems. It’s a reasonable, doable first path that institutions could take right now.
Once you have the detection system running, then you can carefully analyze patterns of failure with an aim to somehow change the system to prevent similar events from happening in the future. We went back and analyzed the causes of these ADEs that we were detecting. The most common cause of injuries, 28%, were allergic reactions or other idiosyncratic reactions. We discovered the patient’s allergy by giving the drug.
Mistaking incidents for errors
Now everybody would agree that those were not errors. There was no reason to not give the drug if the patient had no prior history of an allergy. It was still an injury though. Using the computer, Evans was able to reduce the rate of those idiosyncratic reactions by a factor of about four. When a physician ordered a medication that we knew, historically, had high allergenic potential, and where there was an alternative, less dangerous medication, the computer would automatically remind the physician that there was an alternative drug.
The second most common cause was failure to adjust drug dosage among patients who had declining kidney function. When patients are hospitalized, frequently their kidney function falls; it’s a very commonly observed phenomenon. It doesn’t harm the patient by and large; most have several times the kidney capacity they need. But it does change the pharmacokinetics of drug absorption and excretion, which results in higher peak drug levels.
You can estimate kidney function by estimating the patient’s body surface area from height and weight. Blood creatinine levels, indicating kidney function, are part of a standard chemical panel.
Evans programmed the computer to compare every drug dose we give against estimated kidney function based on height, weight, and blood creatinine levels. From that, it estimates peak blood levels of that particular medication for each patient. Anytime either creatinine levels or peak blood levels are too high or too low, it alerts the care delivery team — in real time — that they ought to consider changing the dosage and it tells what the dosage ought to be.
There are a series of causes: failure to adjust for a patient’s age or simple errors in writing down the order, for example. When we get down to that point, we’re down to 3% to 4% of all ADEs. By the time Evans worked his way down the list, we very significantly reduced the total number of ADEs treated at LDS Hospital.
Q. With the emphasis on computerized health care information systems, why did you study the manual detection system in such depth?
A. While we use the computer at LDS, we were interested in understanding the noncomputerized, manual system because we have 22 other hospitals and not all of them have computer systems. We realize that some of them are so small that it’s likely to be a long time before they get to that level of automation. So we were actively seeking some methods where we could receive a similar benefit without being forced to rely on expensive computerization.
Once you have the detection system running in a noncomputerized form, your next step is to find an intervention that, in a similar way, will stop the harm from occurring. A number of us have tried it, but the person who published the article on it was David Bates, MD, from Brigham and Women’s Hospital in Boston.
He showed that if you identify the high-risk patients, particularly, and then send the pharmacist to round on those patients every day, just to look at their drug orders, they can prevent most of the harm. And when you add the computer system, it took down our ADE rate by a factor of about four. So with that combination of detection and intervention, it’s a quarter of what it used to be.
While I used the example of ADEs, I can give you similar examples of hospital-acquired infections, venous thromboembolism, decubitus ulcers, and blood product transfusions. And we’re barely getting started on strength, agility, and cognition.
The point is that you can really make some difference using common quality improvement tools and by starting out with the concept of patient injury rather than error. It’s so much broader and gives you so much more opportunity.
Nuances of language
When you call it error, it slants the improvement initiatives toward human error. But more of our improvements have come from detecting system failure, not human failure, like reducing the allergy rates with our drugs. Everybody agreed that those weren’t errors. Do you think it mattered to our patients a bit? Does it make them any less dead? So the way you say it is important because it tends to prejudice your thinking.
It’s worth mentioning that you still have to occasionally deal with a bad physician or a bad nurse, but it’s rare. You get so much more leverage by focusing on the injuries.
The final thing to be said is that this work teaches you how poorly we do. We have massive opportunities to improve care for our patients. It’s simply the most important work that any caring clinician can be involved in. It’s worthy of our best time and attention, because we can do much, much better for our patients.
Related reading
1. Classen D, Pestotnik S, Evans R, et al. Adverse drug events in hospitalized patients. JAMA 1997; 277:301-306.
2. Leape LL, Cullen DJ, Clapp MD, et al. Pharmacist participation on physician rounds and adverse drug events in the intensive care unit. JAMA 1999; 282(3):267-270.
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.