Physicians Perform Poorly When Patients Need Special Care
[Writer] This is research news from UIC, the University of Illinois at Chicago.
Today, Dr. Saul Weiner, associate professor of Pediatrics and Internal Medicine talks about a study of physician performance that found patients often receive inappropriate care when their doctors fail to take into account their individual circumstances.
Here’s Professor Weiner:
[Weiner] This study, which was funded by the Veteran’s Administration, was sort of inspired by my interest in looking at more broadly how do we measure the quality of care that physicians provide? Most of the research on physician performance as it’s often called, really looks at whether physicians are following rules and algorithms, and that’s actually fairly straight forward to do. If you know that there’s a particular treatment for a particular disease you can look at the chart and see if the patients for that physician are getting that treatment. Are patients, for example, getting put on beta blockers after they’ve had a heart attack and so forth? The problem, of course, is that taking care of real patients is more than following algorithms.
It involves identifying contextual factors or psychosocial issues in patients’ lives and circumstances that are often essential to providing appropriate care. So for example, the patient has a chronic condition like diabetes and their diabetes is getting out of control and the reason is that their vision is deteriorating and they’re no longer able to give themselves insulin correctly. Or their cognitive abilities are going down and they’re no longer able to understand how to take their medication. Or they’ve lost their health insurance and they can’t afford it anymore. Any of these reasons of course are contextual factors and is a physician simply says, “Oh, your diabetes is out of control. Let’s go up on your diabetes medication,” that obviously is not going to be an appropriate plan of care. It looks good on paper because it’s the appropriate treatment protocol for someone with poorly controlled diabetes but for that patient, at that time, it actually isn’t going to work.
And we’ve come up actually with a term to describe these types of errors. We call them ‘contextual errors’. So to study these types of errors and to see how frequently physicians make them, we hired a team of actors who worked as undercover or unannounced standardized patients. We recruited 100 physicians who were all experienced primary care doctors at a variety of practices in the Chicago and Milwaukee areas. Some of them were VA physicians; some of them were non-VA physicians. They represented the whole gamut of physician practice in this area in primary care, and they generously and kindly agreed to a protocol where some of their patients would be essentially unannounced standardized patients.
They would be actors or fake patients, and they wouldn’t know which ones were in that guise. We trained these actors to do the same things over and over again, to be standardized, and they were prepared with scripts that essentially test the physician’s diagnostic and management skills in a few different ways. Some of these were very straightforward cases where we were just kind of indexing. Does the physician handle an algorithmic, straightforward situation appropriately? And then we added some layers of complexity and variance of these cases. We would add a biomedical layer of complexity where the actor would sometimes have a more complex biomedical presentation, and then in other variance, there were contextual issues that were essential for the physician to identify and address. So for example, a patient may come in—let’s take the asthma instance—with poorly controlled asthma, and the actor would be trained to say something like, “Boy, it’s been tough since I lost my job.” Now, in the baseline version, if the physician said, “Oh wow, what’s going on with your job? Have you lost your health insurance?” the actor would reassure them and say, “No, no, I’m on my wife’s plan,” and then the appropriate care would be simply to go up on the medication for asthma.
In the contextual variant, if the physician were to pick up the actor’s comment, “Boy, it’s been tough since I lost my job,” the actor would say, “Yeah, it’s been really tough. I haven’t been able to afford my medication. I’m stretching it out. I’m taking it every other day,” and so forth, and then we would look to see whether the physician actually picked up on that and addressed the insurance problem. So for example, we would put that actor, or the patient they’re playing on a very expensive designer drug and we would see whether the physician recognized that and put them on a less costly, generic version that the patient could afford or at least refer them to a drug assistance program or something appropriate.
So we actually coded for whether physicians picked up on these red flags, and if they did pick up on them, we also coded for whether they then provided appropriate care.
What we found is that in the baseline versions of cases, physicians correctly addressed the patients’ problems three quarters of the time, roughly 73%. But when there were contextual issues, when it was necessary to pick up on, for example, the loss of health insurance, the performance of physicians dropped very dramatically. Only 22% of the time did they manage those cases without making an error—what we would call a contextual error.
In the biomedically complex cases, they sort of performed in between those two high numbers and low numbers, around 38% where they would get the biomedically complex case correct. In instances where there were both biomedical and contextual issues, they not surprisingly fared most poorly, with only 9% of the physicians getting a case correct in those instances.
So, what we learned from this is that physicians do a pretty good job when there are standards, guidelines and protocols, and when following those standards, guidelines and protocols is all you need to do. But when there are contextual factors that are often critical and in fact ignoring them would be inappropriate, we find that unfortunately only a minority of physicians do pick up on those contextual issues and address them and provide appropriate care.
So, the question of course is, ‘What does this mean?’ I think one of the findings that was striking is that there was no relationship between the amount of time physicians spent with patients and whether they got these contextually complex cases right, so it’s not just a question of they need more time with patients. There’s something different about the physicians who got these cases right in terms of the way they interacted with the patients, the kinds of questions they asked, how they followed up on those questions, whether they prioritized that information, whether they recognized the contextual issues were important, and so it’s probably an issue of physician education. It’s probably fundamentally a question of how do we train physicians so that they don’t over-prioritize biomedical information at the cost of perhaps not paying adequate attention to the contextual issues, which are sometimes equally as important to providing appropriate care.
[Writer] Saul Weiner is an associate professor in Pediatrics and Internal Medicine. For more information about this research, go to www.today.uic.edu, click on News Releases and look for the release dated July 19, 2010.
This has been Research News from UIC, the University of Illinois at Chicago.
