They are quick, convenient and help out colleagues and their patients. What could be wrong with curbside consults?
Quite a lot, according to a recent study. Researchers at Denver Health Medical Center in Colorado recently analyzed curbside consults received on their hospitalist consult service. In the study, one hospitalist would perform a curbside as requested and another would actually go see the patient and officially evaluate them. The doctors' findings and recommendations were then compared. In more than half of the 47 studied cases, the information provided to the curbside consultant turned out to be inaccurate or incomplete, and the hospitalist who visited the bedside made different recommendations than were made in the informal curbside consult.
The results cast doubt on the safety and effectiveness of curbside consults, according to Marisha Burden, MD, ACP Member, hospitalist and lead co-author (along with Ellen Sarcone, MD, ACP Member) of the study published online by the Journal of Hospital Medicine last October. She recently spoke with ACP Hospitalist about how hospitalists should respond.
Q: What led you to have doubts about curbside consults?
A: When you're a resident, or even a young attending, you're sort of thrilled to get curbsides because you have so much work. You try to do anything to help make things more efficient. I used to enjoy getting curbsides, but over the years, sometimes [after the curbside consult] I would peruse through the data or go to the patient's chart and look over things. The more I did that, I started realizing that what I was being told was often not accurate, or there would often be significant omissions of pertinent data. Over the last six and a half years of being a hospitalist, I slowly adapted my own process, which is to minimize curbside consults and to instead offer a formal evaluation of the patient.
Q: What particularly motivated the study?
A: About two years ago, I received a curbside consult on a gentleman with abdominal pain who had been transferred from the medicine service to the psychiatry service. Their question was how to further evaluate the patient's continued pain. They had mentioned that they had spoken with a colleague of mine about the patient's platelets being elevated and that they thought that this was secondary to iron deficiency anemia. Because I don't typically provide curbsides anymore, I went to see the patient, and not only had they curbsided my colleague a few days before, but they had also curbsided the microbiology lab about a possible positive blood culture.
It turned out it was two curbsides gone awry. The patient's elevated platelets weren't secondary to iron deficiency anemia and the blood culture that was thought to be a contaminant was not. Officially seeing the patient made this very clear. The patient had an intra-abdominal abscess with bacteremia. We ended up doing an M&M on that case. My thought was, “Has anybody studied curbsides?” We dug into the literature. There are about 20 to 25 studies that specifically address curbsides, but what we found is they looked at the quantity of curbsides, what specialties receive them, and how much time it takes to do them. The more interesting studies looked at physician perceptions of curbsides. There were concerns about the accuracy of information. I found it really ironic that there's literature stating that physicians are leery of the information, yet we do this on a very frequent basis.
Q: What was the most dangerous curbside incident uncovered in the study?
A: There were 10 major management changes of the 47 [curbsides] that occurred in our study. The most concerning one during the study was a patient on the obstetrical/gynecology service. The information they gave us was, “We have a patient who has been febrile for the past few days. They're on ceftriaxone, they have pyelonephritis and they are mildly hypoxic. We got a chest X-ray and there was a question of atelectasis. The left hemidiaphragm was maybe a little elevated. When should we repeat the X-ray?” That was all the data we were given.
When we followed the protocol of the study, which was to do a formal consult, the patient's blood pressure was 70 systolic, the heart rate was 190 and none of that information had been relayed during the curbside consult. What we found was that the pyelonephritis was secondary to a very resistant type of bacteria and the ceftriaxone was actually not covering the organism. The management plan for this patient dramatically changed. Our original question was “When do we get another X-ray?” and it actually turned to “Let's treat severe sepsis.”
Q: Have the findings changed practice at your hospital?
A: We're in the process of addressing that. We recently had a weeklong improvement event at our hospital that focused on improving the consultation process. We decided that any patient-specific questions should be first vetted through the team's attending and if the patient indeed needed a consult, that there would only be an official consult. Consult teams will not be allowed to convert formal consults to curbside consults. Consultant teams will offer official consults to all curbside consults. It is now expected that any patient-specific questions will be officially seen.
Q: Is there any way hospitalists can improve the quality of the curbsides they provide?
A: That was one of the outcomes we had hoped to figure out: Is there some sort of predictor? If you have to ask a certain number of questions, or if it comes from a housestaff member, is it more inclined to have incorrect or inaccurate data? Ultimately, of all the predictors that we looked at (the team, level of training, whether or not the curbside consultant felt that the curbside was sufficient, the number of questions the consultant had to ask of the requestor, etc.), there was no clear way to predict whether you were going to get good information to make your recommendations from.
Q: Then is the only solution to eliminate curbside consults?
A: That's a tough call. Some recommendations have been made in the past by Manian and Janssen [in a Jan. 10, 1996 study in the Journal of the American Medical Association]. They recommended that if there are two or more confounding variables to a patient's case, you should do an official consult, meaning if it's of sufficient complexity, you should just go see the patient. But I think defining “complex” is very hard, especially if you're not getting the whole true story of a patient's case. In our study, we found that over 50% of our curbsides had incomplete or incorrect information. We also found that our curbside consults were generally complex and most if not all had two or more confounding variables. In over 80% of the curbsides we received during the study, we had to ask three or more questions during the curbside consult to try to answer their question over the phone. It appears that the curbside consults that we as hospitalists receive are somewhat on the complex side, so it's a hard argument with this data to say that curbsides are OK. My own personal belief is that I rarely engage in them, or if I do, I at least preface the conversation [by saying], “I can only offer advice based on what you're telling me.” And now of course, I reference the study.
Q: Short of total elimination, are there any other potential improvement strategies?
A: In our study, we did not allow the person who was receiving the curbside to look up the information in real time. They weren't able to double check the data. In past years, when I was a more junior hospitalist, I would go to the patient's chart or review the data in the computer and the patient's vital signs. I think with today's information technology it is a little easier to review the data from a computer to see if what you are being told is the complete story. However, you are adding to your amount of work, and you are providing a partial consult. I would recommend that providers be leery of the information that they are receiving during a curbside.
Q: Are there any other lessons for hospitalists in your findings?
A: Each institution should take a look at this and determine if they feel this is a safe practice. I think if you were to discuss this with most patients, they probably wouldn't like the potential ramifications of a physician who has not seen them or officially reviewed their data making recommendations on their care. Given the fact that 51% of our curbside consults in the study had inaccurate or incomplete information and that there were subsequent management changes, some major, there's considerable risk for a physician or any provider that the recommendations you are making may not be correct based upon what is really going on with the patient. And ultimately there is the risk that the patient's care may be compromised.