Patients Go Undercover to Record Encounters with Doctors

UIC Podcast
UIC Podcast
Patients Go Undercover to Record Encounters with Doctors

News Release


[Writer] This is research news from U-I-C — the University of Illinois at Chicago.

Today, Dr. Saul Weiner, professor of medicine, pediatrics and medical education at UIC and health services researcher at Jesse Brown VA Medical Center, talks about his recent study on the importance of patient-centered decision making to improve health care outcomes.

Here’s professor Weiner:

[Weiner] My name is Saul J. Weiner, I’m a physician and health services researcher based at Jesse Brown VA Medical Center and I’m also a professor of medicine, pediatrics and medical education at the University of Illinois at Chicago. In addition I hold a appointment as deputy director of the Center for the Management of Complex Chronic Care, which is a VA Research Center of Excellence.

This project was the most recent in a series of studies that my team has been doing for a number of years. Our interest has really been in looking at the relationship between physicians addressing contextual issues and patient outcomes.

The project really dates back to my own experience supervising residents and finding that they were really getting pretty good at following algorithms and addressing best practices, but that often times there were circumstantial issues in patients lives – their preferences, and needs, and so forth – that were pretty essential to identify and address in order to come up with a good care plan. And often times physicians would essentially miss these issues or maybe notice or not recognize that they were critical to the care plan.

And so that led us to a large study we did a few years ago, that was also published in Annals of Internal Medicine, where we actually hired some actors to work as undercover patients. We called them unannounced standardized patients. And we set up scripts where they would essentially have to role play patients who had complex psycho-social issues and they would drop clues. So for example, they might come in and say it’s been tough since they lost their job and they would have a chronic condition that went out of control and the underlying problem, of course, is that they couldn’t afford their medication anymore. We would look to see if the physician picked up on that, asked about it. If they did, we would have the actor reveal that in fact they were not able to afford their medication. And that, of course, would lead to the physician, hopefully, recognizing that they were on an expensive brand name drug, putting them on a generic, and so forth.

If the physician kind of missed the boat the patient would usually be sent out on a higher dose of a drug that they already couldn’t afford and weren’t taking.

So we became very interested in this phenomenon. Though that paper showed that there was a high tendency for physicians to overlook context even when we had actors really planting clues for them.

The purpose of this study was to involve real patients and to look at what happens in real encounters. So this was pretty challenging because we had to enlist hundreds of patients to audio record their encounters with their physicians undercover. And of course we had to find physicians who would allow us to follow this protocol. And physicians were very gracious in willingly allowing us to have their patients audio record encounters at times when they didn’t know they were being audio recorded.

The other challenge, of course, was that we never knew what was going to happen. We didn’t plant clues. We had to see if in fact in these encounters, after the fact, the patient presented with real psycho-social issues which were essential to address.

One of the things we did after an audio recording was accomplished is we went and looked at the medical record and had a very specific scoring system for looking for contextual issues. So, for example, if a patient had missed a lot of appointments we would consider that a red flag. In other words, it’s something the physician should be asking about because it’s clearly interfering with the patient’s care.

Conversely, if the patient has for example a chronic condition like diabetes or hypertension that’s going out of control we would say that that’s also a contextual issue. That’s probably a sign that something is going on in that patient’s life that needs to be addressed.

So we would then listen to the audio recordings and see whether the physician noticed these problems and asked about them. Or did they simply put the patient on a higher dose of the insulin or the blood pressure medicine they were taking. And, of course, that would probably be unsatisfactory if the underlying problem wasn’t discovered and addressed.
But one of the challenges of dealing with real patients is that we wanted this to be as scientific as possible. In the first study with the fake patients we actually designed the cases, we knew in advance what would be good care and what would be missing the boat.

In this situation we had to design a coding system that had high inter-rater agreement. In other words, we wanted a system where multiple coders could listen to an audio recording, independently come to the same consensus about several things: number one, was there a red flag and did the physician notice it? Number two, if they noticed it did the patient reveal some issue or factor that was clearly relevant to their underlying presenting problem; and three, if that occurred, did the physician then address it in their care plan?

And what we found is that there was very high inter-rater agreement in terms of whether different listeners and coders would agree on whether the physician had done what they were supposed to do.

Now the key part of this was that we then had a group of coders who followed these patients for nine months to see how they did.

So for example, if the patient came in with a poorly controlled chronic condition we looked to see whether that condition improved.

If they’d been missing a lot of appointments we looked to see whether they stopped missing appointments.
After that occurred then a statistician essentially put the two data sets together to see whether those physicians, those encounters who addressed contextual issues, in those situations, were those patients more likely to do better. Were they more likely to have the underlying, red flag issue resolved.

And what we found was that there was very much an association between situations where physicians picked up on these contextual issues and how the patients did during the subsequent nine months in terms of that very specific issue.
But I think what was intriguing about our study was that we were actually able to show that that relationship is real and I don’t think that’s been shown before. And I think what it really tells us is that the information that patients divulge during appointments about their life situation is critical to address and to take into account if we’re looking for optimal health care outcomes.

In this study when we had real patients, first of all, in about two-thirds of the encounters there were contextual factors that should be addressed.

In 59 percent of those situations where there were contextual factors that were essential to the care plan, the physician actually made a contextualized care plan and in 41 percent of the time they didn’t.

And then we also looked at outcomes in those situations and, of those, 59 percent of cases where the physician made a contextualized care plan there was a good outcome in 71 percent of those cases and a bad outcome in 29 percent. And you can compare that to the 41 percent of the situations where physicians did not come up with a contextualized care plan, in those situations a good outcome only occurred in 46 percent of cases and a bad outcome occurred in 54 percent of cases. And when the statisticians run those numbers what they find is that those are statistically significant, meaning that there’s a 95 percent or greater probability that that relationship is not due to chance.

[Writer] Dr. Saul Weiner is a professor in medicine, pediatrics and medical education.

For more information about this research, go to www-dot-news-dot-uic-dot-edu ( click on “news releases” and look for the release dated April 15, 2013.

This has been research news from U-I-C — the University of Illinois at Chicago.

Print Friendly, PDF & Email