This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Medical Education, is properly cited. The complete bibliographic information, a link to the original publication on http://mededu.jmir.org/, as well as this copyright and license information must be included.
Collaborative reasoning occurs in clinical practice but is rarely developed during education. The computerized virtual patient (VP) cases allow for a stepwise exploration of cases and thus stimulate active learning. Peer settings during VP sessions are believed to have benefits in terms of reasoning but have received scant attention in the literature.
The objective of this study was to thoroughly investigate interactions during medical students’ clinical reasoning in two-party VP settings.
An in-depth exploration of students’ interactions in dyad settings of VP sessions was performed. For this purpose, two prerecorded VP sessions lasting 1 hour each were observed, transcribed in full, and analyzed. The transcriptions were analyzed using thematic analysis, and short clips from the videos were selected for subsequent analysis in relation to clinical reasoning and clinical aspects.
Four categories of interactions were identified: (1)
The dyad VP setting is conducive to activities that promote analytic clinical reasoning. In this setting, components such as peer interaction, access to different resources, and reduced time constraints provided a productive situation in which the students pursued different lines of reasoning.
In professional education, students need to apply facts and concepts into relevant work-life situations. For medical students, it can be challenging to apply biomedical knowledge when entering into clinical practice; this application has previously been described as “slow, awkward, or absent” [
Two models are commonly used to describe the nature of clinical reasoning processes. The hypothetico-deductive model describes reasoning as starting with the generation of hypotheses, followed by analytic evaluation of these hypotheses [
An interactive virtual patient (VP) allows students to gather information in a stepwise manner and suggest diagnosis and management. Relevant VP cases have been shown to engage students in active thinking and decision making [
The use of computer applications in small group settings has generally been shown to be beneficial for learning [
An exploratory observation was conducted to identify interactions and delineate their characteristics during VP case sessions performed by students in dyads. The students were third-year medical students during their clinical rotation at the Rheumatology Unit at the Karolinska University Hospital in 2011. Four VP cases constituted a mandatory task, which was recommended to be conducted in pairs. The VP assignment was not scheduled at a specific time or graded, but it served as a basis for discussion with a clinical supervisor at the end of the rotation. The VPs were based on authentic patients and authored in a derivative platform of the NUDOV system described in Wahlgren et al [
The construction of themes was data-driven, that is, not directed by a priori categories. The first session (session #1) was transcribed in full and a preliminary thematic analysis was performed [
Ethical approval was granted by the regional Ethics Review Board in Stockholm, Sweden (#2009/609-32/5).
Four categories of interactions related to learning were identified: (1) task-related dialogue, (2) case-related insights and perspectives, (3) clinical reasoning interactions, and (4) interactions with other resources. Each category is presented below, and interactions from the different categories are illustrated using quotes from the two sessions, indicated by the session number and point in time of the respective session.
Part of the dialogue was dedicated to understanding how to approach the task and navigate in the software. Interactions in this category were related to, for example, the students’ perceptions on how the assignment would be followed up by their supervisors, and, more directly, how the interface worked, in particular where they had to click to navigate in the VP software:
Yes, exactly. May I just ask: what are we going to report on Friday? It was going to be about the diagnosis we identified and then about the management, right?
Exactly.
OK, I just wanted to be sure.
But, by the way, did we have any...It says 3 questions, but we could ask many questions, couldn’t we?
Yes, it...
Whether he is on any medication, perhaps?
Yes, but the question is...do you think it will register it?
No, but if we just imagined.
Yes, we can make up questions.
I was just thinking...should we click through step by step? Or we can just adopt a specific approach.
Yes, perhaps we should just choose what’s relevant and then look at things again later.
But for now we’ll follow these ones, anyway.
Yes...
Since only one person could select and write text into the software, there were negotiations about control, for example, what to select and what questions to put to the patient. Sometimes, the pairs divided tasks between themselves. For example, one could read on the screen while the other one looked up facts in a textbook:
It might be far-fetched, but...should I read all blood test results?
Here: it should be below, or between 60 and 400, but go ahead and read. Keep reading.
I’ll read quietly, so that you can check there.
OK.
What more do we want to know? For the purposes of our own learning...We want to know more...about how the pain varies?
Yes, and when. Whether he suffers from morning stiffness, whether motion relieves it and whether rest worsens the pain.
I want to know if he is affected in any other way, if he has any other symptoms, any other...
Through interaction with the VP cases, the students obtained insights related to symptoms and diagnoses and identified new clinical perspectives. The process of identifying differential diagnoses and the progress toward the final diagnosis generated discussions and reflections, based on information obtained from the patient and clinical findings. The software and the way the VP cases were constructed allowed a free flow of ideas, several of which were followed up at later stages. Students reflected upon differences between this setting and authentic patient encounters, which they perceived as more constrained because of time restraints. They referred to previous experiences of feeling pressured to appear as if they already had knowledge in front of patients and supervisors:
See how much we are able to think about when sitting like this. When you are with a patient so ehh...
It’s because you don’t have a lot of time. You have to focus on behaving properly in front of the patient and so on.
There was plenty of time to elaborate on findings. Ideas and hunches could be followed up. The clinical information in the VP case was presented in a variety of ways namely, in text, short video clips of the patient answering questions, or filmed examination procedures. In session #2, the students were inspired to try out a practice physical examination on themselves while watching the procedure, and they watched the procedure one more time after that:
Well, let’s check him over, okay?
Yes, I agree.
Then, it should normally move like this, right?
May I try it on you?
Somewhere here?
Yes.
Schober’s test.
So he is just bending the hip joints, not like that.
Yes, exactly.
Exactly, not like this when he also is bending the back.
Can we look at it again? I would like to see it one more time.
In one case, the reference of nonsteroidal anti-inflammatory drugs (NSAIDs) led to a discussion about the mechanism of action of such drugs and their effects in relation to the assumed diagnosis:
Yes, because it is an NSAID.
Yes, exactly.
But, oh my god, isn’t this weird? Well, NSAIDs dampen the inflammation, but we want to prevent even more, don’t we?
So you want to give corticosteroids, or something like that?
Yes, or wait, was he still on that? The white blood cell count was high as was the sedimentation rate...We don’t want to just stop the symptoms, right? We want to stop the progression, don’t we? If you understand what I mean...
Yes, I understand, but NSAIDs are anti-inflammatory drugs, this is what they do.
Well, yes, but it’s only COX that is inhibited.
Yes, it’s quite a weak inhibition of the inflammation one could say.
Then it’s only leukotrienes and prostaglandins that are not being produced, which means that you don’t see...
Leukotrienes are produced, it’s the prostaglandins [that are not produced], and thromboxanes are not produced either.
Correct, it’s only those that are not produced.
But still, it [the drug] inhibits quite a lot of the inflammation, one could say.
Yes, that’s true.
But, yes, it’s not the same thing as corticosteroids or methotrexate, or things like that.
It does not inhibit the lymphocytes per se, even if not as many of them stream out, maybe.
Yes, you are not supposed to see the same upsurge.
But again, we want to prevent something severe here so that it doesn’t result in a bamboo spine, which is permanent.
Yes, I understand what you mean, but at the same time I think that there must be a reason why they treat it this way. But we could read more about it maybe. It is also stated here that continuous physical exercise prevents worsening of the functional status, so I note this recommendation: exercise!
The VP cases were based on authentic patients; they were therefore not textbook examples. Students identified inconsistencies in relation to classification criteria; yet, their suggestion of diagnosis at different stages during the session—rheumatoid arthritis (RA) in the following instance—was also based on references from real-life clinical complexity:
I thought of RA.
I also thought of that.
But isn’t it small joints that are affected first?
Yes, I also thought about that, and he mostly seems to have problems with large joints.
But we could note it down as an alternative diagnosis. Not everything has to be according to the textbooks.
Yes, true.
The students brought previous knowledge into the reasoning. However, in many instances, the level of knowledge varied between the 2 individuals, and on several occasions it was incomplete. They supported each other by filling knowledge gaps and looked up information when they were uncertain:
This sounds pretty much like Bechterew to me.
I mean, I don’t remember so much about that disease. Could you remind me?
Sacroiliitis, and some other things...
I always confuse CRP and ESR, which is which...One is supposed to be below 100, and the other one below 3...I think.
Yes, but the one with 100 – it’s only when you have a bacterial infection that it can be over 100.
That was it. CRP, right?
The clinical reasoning interactions consisted of uncertainty, questioning, clarifying, and verifying dialogues. The dyad setting encouraged the students to generate explicit hypotheses, as well as to proceed with confirming or rejecting these hypotheses. In both sessions, there was often one and the same peer giving suggestions to further advance the reasoning:
OK, then I think we can decide that he most probably has PMR, and so I think we initiate him with glucocorticoids to see whether he gets better. Because this way we can confirm the diagnosis, right?
Yes, sure.
The following quote provides another example of how the reasoning is verbalized and the thread of the reasoning is made explicit. The students updated themselves on diagnostic criteria, and this information guided both their focus when further interviewing the patient and their interpretation of the patient’s answers and other findings. The reasoning revolved around symmetry and the patient’s ways to express the location of pain.
Should we consider polymyalgia rheumatica?
Yes, I am not sure what...
I will check what it is.
Okay. But here it is stated that chronic idiopathic myositis can be an isolated inflammatory systemic disease, or part of another rheumatic disease such as Sjögren’s syndrome, systemic sclerosis, mixed connective tissue disease, systemic lupus erythematosus or rheumatoid arthritis.
Yes...
And then with regard to polymyositis in particular, it is stated that the predominant symptoms are decreased muscle strength, decreased stamina in the proximal muscles, shoulders, nape of the neck, thighs, and the pelvis. Symmetric distribution. So we need to know whether it is symmetric. Myalgia may occur, but it is not as common as the weakness.
Was there no question about the symmetry?
No, not really.
And then it is stated that acute myositis, which mostly is seen in conjunction with viral infections, is established quicker and is often followed by myalgia. So it could be this.
I’m sorry, what did you say?
Acute myositis, which is mostly seen in conjunction with viral infections, is established quicker and is often followed by myalgia. So it could definitely be this, too.
That's true. And this was that...poly- and dermatomyositis?
Yes, exactly.
And he responds quite inadequately to the question about symmetry: “It is located in the shoulders and hips.” So it should be that...Because otherwise, he would say: “the right shoulder.” [The student pats her shoulder] and...
Hm, yes, it should be that.
And here, for polymyalgia rheumatica it is stated that [the student reads from the textbook] “new onset of relatively acute established mechanic pain in the proximal parts of the arms, shoulders, nape of the neck, and/or hip areas and thighs is characteristic of PMR, as it is abbreviated. Patients describe intense morning stiffness, difficulties turning over in bed, getting up out of bed, and putting on clothes during the morning hours.”
Okay.
In general, the symptoms are fully developed within several days to a couple of weeks. Constitutional symptoms, such as fatigue, subfebrility, loss of appetite, and weight loss...
Well...hehe.
It’s crystal clear!
Furthermore, the students verbalized interpretations of radiographic images. In session #2, radiographic visualization of the spine evoked interest, and the peers helped each other to understand the findings and relate them to the patient’s symptoms. The software displayed the images, but there were no indications of what to look for (eg, arrows), or how to interpret the findings. The students realized that they could not fully interpret one of the images:
Yes. Here, we can see a little better. Let’s see. Here, it is very uneven; and here, it feels like it starts to become more linear.
Hm, there they are evened out. They are evened out there.
These ones stick out like that...
And they are also evened out, I think.
Yes, exactly.
Here, however, you can see that it’s fine.
Still evened out, that is what I see.
Yes.
Yes, now we are up there.
Yes, the cervical spine.
This one is very difficult to interpret, I think.
I find it absurdly difficult, too. Yes.
Has it grown together here? And here, maybe? Here too. I don’t think I am competent enough to interpret that one, actually.
No, that’s true. I’m not competent enough for that one either. I mean, I’m not saying that you are not competent; I’m just saying that I’m not competent.
The students worked on the VP cases in conjunction with other clinical tasks during their clinical rotation. Several times, they referred to patients and procedures they had seen before. Examples from real-life experiences at the clinic were used to illustrate representative instances in the VP cases and facilitated clarifications during reasoning:
Is it possible to have psoriatic arthritis without the typical skin lesions?
I think it is, but I think it most commonly affects the DIP [distal interphalangeal] joints.
The psoriatic arthritis?
Yes, or am I wrong?
Yes, they are the ones most commonly affected, but it can also be...I mean...the man I saw earlier today, he had...
Yes?
Well, I was not responsible for him on my own, but I talked to him for a while. Both of his wrists were swollen, here, and here, and he had pain in one shoulder, an elbow, and in both of his feet; so it was quite extensive.
Yes. Did he have any other symptoms?
No, those specific joints were swollen and tender.
Resources other than the VP platform were also used; mostly a textbook but also Web-based medical resources, lecture notes, and a list of laboratory tests. The students were allowed to use other resources, and they did so when they found it helpful in making the diagnosis, when they wanted to relate content in the VP case to classification criteria, guidelines, and common management routines, and when identifying knowledge gaps:
They were like evened-up corners of the vertebrae, I think.
Yes, I remember that. But what is the source of the pain? I mean, what’s happening? What’s the reason for the patient experiencing pain? Well, I think I should read a little about it in the textbook. I’ll look it up. [The student opens the textbook].
Well, yes, we can have a look; it’s something we should know anyway.
This is, in fact, an awesome way to learn!
Yes, definitely.
Especially when you have such a good textbook.
Yes, it’s actually a really good one.
Imagine if we had such a good textbook during all rotations.
It’s actually extremely useful to work in a problem-based manner sometimes.
Hm, okay. But wait...should we look at...what's that? Is the ESR [erythrocyte sedimentation rate] elevated?
55.
And it is supposed to be below 3, isn’t it?
Yes, normally yes...
Well...we should look it up.
Yes.
Okay.
I will just check.
Ah...
My goodness, that was very slow...
My son had an ESR of 29"...Okay, it is supposed to be below 8.
In this study, we identified and described interactions within student dyads in VP-based learning sessions. Our observations revealed elaborate reasoning processes supporting the development of analytic clinical reasoning. Overall, the dyad peer setting contributed to fruitful interactions and promoted the development of analytic expertise.
The assignment that framed the students’ task was loosely regulated by the teachers, and the VP interface allowed for relatively free exploration of the patient cases, which made the task-related dialogue between the students necessary. During the task-related dialogue, a shared understanding was created on how one should approach the task to gather patient data in the case-related context.
Diagnosis-specific facts were elaborated upon, reference values from previous experience and other resources were used, and a variety of procedures were observed and discussed. The evaluation of key findings had positive consequences and resulted in structured gathering of further information and suggestions for managing the respective patient. To some extent, the knowledge was already present and readily available in the students’ reasoning. However, in several cases, the students had to search for information or ask each other. The verbalization and application of knowledge seemed to add further value to preexisting knowledge, since it was put into a clinical context. Biomedical facts were thus interwoven with the clinical case in a very active manner, connecting knowledge and procedures in a meaningful way.
The verbal interaction between students made it possible to elicit reasoning processes that otherwise might have remained implicit or would not have occurred. Some of the findings pertain to a specific case while others are more general, for example, processes related to the development of reasoning strategies or decision making.
Previous literature on clinical reasoning is generally characterized by categorization into hypothetico-deductive analytic approaches and experiential-based nonanalytic approaches [
The setting in our study was based on VPs, using a structure that supported making a diagnosis and suggesting appropriate management of the patient before the real-life outcome of the respective patient was revealed. The students’ interactions were clearly driven by the VP design, and the specific cases were always the central focus. Nevertheless, the students used several resources other than the VP software. A textbook, lecture notes, and the Internet were utilized to gather information that provided evidence needed for further reasoning and consolidation of knowledge. The actions within the learning activity were therefore not exclusively directed by the VP software. It is reasonable to assume that large variations in interactions were influenced by the possibility of accessing various other resources, as well as by the design of the VP scenario [
Our observations suggest that the peer setting in dyads was pivotal for the elaborate reasoning observed, as it provided constructive resistance while processing the VPs. It is reasonable to assume that part of this reasoning could have occurred internally within an individual in a nonpeer setting; however, the reasoning would not have been explicitly voiced; it would have occurred in silence, and it would certainly not have been critiqued or evaluated by a peer. Experimental designs have found that group cognition differs from individual cognition and that the dyad setting is productive for abstractions [
The constructive resistance that forms part of peer contribution may help to adjust topics to individuals’ level of knowledge and experience. Still, a concern that has been raised is the risk of knowledge imbalance within the dyad, or incompatible pairs [
The arranged VP situation allowed the pair to take the time to reason broadly, ponder over certain issues, and even experiment with physical examination procedures. This slowed-down, broader reflection is rare during a time-constrained real-life patient encounter. The VP setting provides a less stressful milieu for reasoning and reduces the perceived demands on having to appear knowledgeable in front of patients and educators [
A major strength of this study was the in-depth analysis of the sessions. However, the low number of sessions studied could be considered a limitation, as it limits the generalizability of our observations when applying them to other contexts. Thus, the identified themes are neither exhaustive nor expected to be replicated in all settings. Moreover, the characteristics of the specific cases and the VP interface may have influenced the reasoning process, and therefore the findings cannot be generalized to any VP dyad setting. The methods of observation provided not only an increased awareness of how students’ reasoning processes function while exploring a VP case but also an insight into how peer interactions may relate to clinical reasoning. However, it is worth noting that the students were aware of the fact that they were being filmed during the sessions, and a social desirability effect may have impacted their interactions.
By actively taking part in verbal reasoning and highlighting the process as well as the outcome, the students increase the meta-awareness of their reasoning [
The number of VPs in this setting was small. Therefore, they do not substantially contribute to the nonanalytic aspect of expert reasoning. However, the dyad setting contributes to an in-depth encounter with a few cases supporting the analytic thinking. This analytic thinking process, and the awareness thereof, requires cognitive efforts and time to be processed. VP scenarios and the surrounding educational framework should therefore be designed carefully, taking the benefits of reasoning into consideration [
VP activities can be organized in different ways; these different arrangements may influence both the learning process and learning outcomes [
The dyad setting enabled a stepwise investigation of VP cases and elaborated reasoning. Both analytic and nonanalytic reasoning occurred during the interactions. The VP activity also triggered interactions with other sources, which served as tools for information gathering and contributed to consolidation of knowledge. The VP design and the dyad arrangement enabled reasoning and rigorous learning processes that are unlikely to occur in individual settings.
Supplementary methods information.
nonsteroidal anti-inflammatory drug
rheumatoid arthritis
virtual patient
None declared.