Conversations on Diagnostic Uncertainty and Its Management Among Pediatric Acute Care Physicians

We conducted a qualitative study through focus groups with PEM and PHM physicians. This study informed a later quality improvement project to identify pediatric inpatients with diagnostic uncertainty.¹⁹  Focus groups occurred between May 2018 and August 2018. Physicians were separated into focus groups by specialty. Approval from our institutional review board was obtained before beginning the project.

The study setting was a large quaternary care academic pediatric medical center with 692 beds. The medical center includes a free-standing satellite campus with its own emergency department (ED) and 42-bed inpatient unit.

Participants were attending PEM and PHM physicians who care for patients at both the main and satellite campuses. Participants were volunteers recruited through E-mail invitation. Focus groups were held at times convenient for physicians to accommodate clinical schedules. Participants were not compensated for their participation.

The interview guide was developed by the authors (A.I. and S.P.) through conversations with content experts and review of the existing literature. Questions were reviewed by the authors and the QI study team and modified for clarity and completeness. Verbal consent was obtained after an explanation of the study was provided by the focus group moderator.

Four focus groups of 3 to 6 participants lasting 60 to 90 minutes each took place in private conference rooms. Focus groups were conducted because authors desired to explore how participant responses built on other participants’ ideas on diagnostic uncertainty. Four focus groups were completed as little new information was generated after the last group. Each group was moderated by a team member with training in qualitative research (S.P.). The interviews were recorded, deidentified, and transcribed verbatim by a team member (S.P.).

A total of 13 physicians participated in the focus groups; 4 participants were male, and 6 participants identified as PEM physicians. Physicians were faculty, and the range of practice years varied from 1 to 28 years.

Transcripts were analyzed in an iterative process using thematic analysis with a constructivist paradigm. Thematic analysis has been widely used in qualitative research as a method for identifying, analyzing, organizing, describing, and reporting themes found within a data set.²⁰  We used a deductive approach, organizing themes and subthemes around the diagnostic process model from the Academy of Medicine report. The primary researchers (A.I. and S.P.) individually reviewed the transcripts, highlighted ideas, and organized these ideas into themes. The themes were compared and discussed until the investigators were in agreement.

Interviews with 20 participants generated 15 themes, which were categorized by using the model of “the diagnostic process” from the National Academy of Medicine (Fig 1.)¹  The themes and exemplary quotations are listed in Table 1.

During the information gathering phase of diagnosis, clinicians noted red flags, which increased the risk associated with diagnostic uncertainty; specifically, multiple providers mentioned that incomplete immunization status of the patient made them more concerned about missing a disease process.

Physicians noted that “gut feelings” and cognitive biases affected their information integration and interpretation. Gut feelings were often described as “something weird is going on” or knowing that “something is not right.” One participant stated, “there’s no term in my head…it’s just a patient whom I don’t know what is going on, but I know something is wrong, kind of like a gut feeling you have” (PHM). Focus groups also identified an awareness that cognitive biases might predispose to diagnostic error. Specifically, clinicians mentioned anchoring and premature closure: “leading yourself down a path…you’re so sure of that’s what it is” (PEM).

Clinicians reported that they employ several strategies to combat diagnostic uncertainty in the information integration and interpretation phase. Participants stated that diagnostic uncertainty caused them to go back to the “drawing board,” and to “consciously go over it all with a blank mind in order to get my own thoughts.” This theme was sometimes linked with repeated information gathering: “I know that I fine tooth comb the chart of a patient that I don’t know what is going on; I’m looking back at old notes… every single weight …every single vital sign point…going through the process of being really systematic with my thought process, I’m absolutely digging into the chart a lot more than I would” (PHM).

Another theme that emerged to combat diagnostic uncertainty was using fresh eyes, or asking a colleague or consultant to assist in the diagnosis. Many participants felt that shift work was particularly helpful in this regard, because handoffs between providers at change of shift are a natural opportunity to get a second opinion from another clinician. “I feel like it was very common to get someone saying I have no idea what’s going on, if you want to take a look at this kid’s chart that would be great” (PHM).

When unable to identify a working diagnosis, physicians noted they “rule out” potential diagnoses that carry significant morbidity or mortality, also known as cannot miss diagnoses. One participant stated, “I think for me, with diagnostic uncertainty, I get to a certain point where this is a list of things that I want to make sure that I’m not missing” (PEM).

Participants noted several ways that they modify their communication with patients and other clinicians in the face of diagnostic uncertainty. In handoffs with other clinicians, transparency about diagnostic uncertainty allowed the receiving clinician to have “their radars up about certain things.” Conversations with the patient’s primary care physician (PCP) were noted to be especially important to relay diagnostic uncertainty and establish a follow-up plan. A clinician stated “I’ll call the PCP myself on this one, just because I want to make sure there is a good plan, that we’re talking about the same thing, on the same road and things are going to go as smoothly as possible” (PHM).

Communication with patients and their families regarding diagnostic uncertainty was not always as direct. Many clinicians would walk through the presenting symptoms, their thought process, and the current work-up for their patient rather than stating the diagnostic uncertainty outright. One clinician described, “I don’t know if I used the word uncertain though…the patient came in with X, Y, and Z symptoms, all of these labs have been positive or negative, and we still don’t have a clear diagnosis, but now that I’m thinking about it, I don’t think I actually say the word uncertain” (PEM).

In the context of diagnostic uncertainty, clinicians thought about treatment through several cognitive and management tasks that involved incremental testing over time and considered the safest treatment location for the patient. Contingency planning, or stepwise diagnostic and treatment decisions based on how the patient responded, was a commonly identified strategy for treating patients with diagnostic uncertainty. “I’d say I have a stronger contingency planning for patients with diagnostic uncertainty. So, with the uncertain diagnosis, I probably have more conversations with residents, or more conversations with my colleagues about it if this is uncertain” (PHM).

A subtheme that emerged among PEM physicians was disposition over diagnosis, meaning that determining the appropriate admission or follow-up plan for a patient in the emergency department was more important than making a definitive diagnosis. A clinician stated, “I just need to figure out is it safe enough to send them home [to follow up] with their doctor or does it have to be figured out [now]” (PEM).

The test of time is another management plan used in the face of diagnostic uncertainty, especially during a hospitalization. Clinicians felt that some patients would declare themselves with time and that a diagnosis did not need to be made immediately, as illustrated by “there’s time, there’s monitoring, there’s doing other things and seeing how a patient progresses” PEM.

Finally, many physicians felt that the resources available in the outpatient setting, such as an established PCP or access to testing, determined the management of a patient with diagnostic uncertainty. One physician stated, “well, if they don’t have follow-up, that gives me a lot of discomfort and that will sometimes lead me to admit when I think they’re probably ok to go home, but if they don’t have a PCP, that gives me a lot of discomfort…then I’m like, where are they going to follow-up? (PEM).

Physicians identified a diverse set of outcomes related to diagnostic uncertainty including clinician response to uncertainty and patient trust with providers. One system outcome that emerged from focus groups was the effect of diagnostic uncertainty on clinicians. In general, participants felt very comfortable with uncertainty and thought it was inherent within the field of pediatrics. “I think that there’s a very, very comfortable uncertainty within pediatrics, and I don’t feel the need to be certain about many things” (PEM). At the same time, diagnostic uncertainty was associated with a feeling of frustration: “uncertain diagnoses make you frustrated because you’ve tried like 18 things…and you’re like what the heck is it? It’s frustrating because you are trying to figure stuff out” (PHM). Participants noted that clinical experience plays a role in physicians’ comfort with uncertainty. The more patient care that the physicians had experienced, the more comfortable they were with diagnostic uncertainty: “As we become more experienced in this [clinical care], you see that it [diagnostic uncertainty] occurs a lot more than it doesn’t” (PHM).

Another outcome of the diagnostic process is the patient’s experience, and diagnostic uncertainty may affect the trust of the patient in the medical system. Several physicians worried that expressing diagnostic uncertainty would cause parents to lose trust in them. One physician stated “I don’t want them [parents] to think I’m dumb or that I don’t know what’s going on or that we aren’t doing anything…I want them to know that we’re working on it, and we are thinking about a lot of things” (PHM). Another commented “I usually don’t use that word [uncertain]… [Instead, I use] something like we are still working on figuring out the explanation, kind of conveying the process and this is what we think is going on, but we’re looking into some other things as well” (PHM).

In our qualitative study on the attitudes and behaviors of pediatric emergency medicine physicians and pediatric hospitalists regarding diagnostic uncertainty, we found several themes which have been seen by others. Physicians described strategies of eliminating feared cannot miss diagnoses, using the test of time and contingency planning. These approaches have been noted by Hewson et al, who set out to identify the “tacit knowledge,” which physicians use to manage diagnostic uncertainty from standardized patient encounters.¹²  Meyer et al also identified the theme of cannot miss diagnoses, and contingency planning was a key strategy noted in interviews of emergency medicine physicians¹⁶  and has been recommended as a good practice by several others.^21,22  Finally, the use of “gut feelings” as a marker for diagnostic uncertainty has been proposed frequently, and we were not surprised to hear it echoed in our study.^13,16,23 

Physician participants endorsed using a diagnostic pause, a cognitive strategy which aims to move the diagnostician from so-called fast to slow thinking when a patient’s presentation was not clear. Fast (type 1 processing) thinking is described as intuitive and involves using heuristics to make a direct association between new information and a similar example in one’s memory.^24,25  Slow (type 2 processing) thinking is thought to be abstract and analytical, consistent with logical rules, and is thought to use working memory.^24,25  Our participants’ preference for a methodical review of a patient’s presentation in the face of diagnostic uncertainty implies their conviction that type 2 processing is more likely to be accurate or correct. This belief has been shared in the literature, with several experts theorizing that abbreviated decision-making, or heuristics, are prone to error and that this risk could be mitigated by deliberately moving from type 1 to type 2 processing.^22,26  After all, the Institute of Medicine report states, “When a heuristic fails, it is referred to as a cognitive bias.”¹  Importantly, the hypothesis that fast thinking increases risk of diagnostic errors and that moving to slow thinking decreases this risk has a mixed evidence base and is the source of substantial debate among leaders in the field of diagnostic safety.²⁴  Some believe that errors arise from both processes.²⁵  For example, 1 study showed that a correct diagnosis was associated with less time spent on a diagnostic task,²⁶  and time did not affect diagnostic accuracy, suggesting that both type 1 and type 2 processing have roles in high-quality diagnostic thinking.^27–29  Finally, others argue that the delineation between fast and slow thinking is not so clear, that physicians routinely use both strategies simultaneously, and that one cannot consciously control one’s thinking to perform fast or slowly.³⁰  Further research is needed in the cognitive processing of physicians and how to improve their accuracy.³¹ 

Many studies on diagnostic uncertainty have taken place in the outpatient setting, so our themes specific to hospital-based medical practice are noteworthy. The importance of disposition over diagnosis is especially important in this setting, because a patient may be safe for discharge before a diagnosis can be made. This phenomenon has been described before in the context of febrile children being evaluated by PCPs who “place less importance on making a definitive diagnosis than on discriminating between self-limiting and potentially serious illness.” An important feature in determining safe disposition included assessment of appropriate outpatient resources, including a PCP. Finally, fresh eyes are a resource that may be more available in acute care compared to the outpatient setting. New providers at change-of-shift and in-house subspecialty consultants are available not only to confer with the physician struggling with diagnostic uncertainty but also to meet and examine the patient in person. The strategy of fresh eyes has been recommended by others and was also noted in the qualitative study by Meyer et al, which included inpatient providers.^17,32,33 

Whereas communication of the diagnosis and a management plan to patients and their families is critical for quality care and shared decision-making, the process is understandably more difficult when there is diagnostic uncertainty. For example, a recent study of emergency medicine residents found a desire for more training in communicating with patients about diagnostic uncertainty.³⁴  Various studies have documented that physician factors, including length of experience and physician specialty, alter how much information is shared with patients.^4,5,17,35  In our study, physicians perceived that they became more comfortable with uncertainty with experience. Our physicians who had practiced for a longer period voiced more comfort with uncertainty than those who had just completed residency or fellowship. Previous studies have documented that pediatricians have moderate levels of tolerance for ambiguity and are more likely to recommend full disclosure of their uncertainty to the patient and their family compared to internists.^35,6  Our physicians expressed the importance of communication but shared reluctance to specifically state uncertainty because of concern that this would decrease trust in the physician. This is not a surprising concern; a systematic review found some patients expressed dissatisfaction when physicians disclosed uncertainty.³⁶  Because uncertainty is common in pediatrics, there may be value in developing best practices regarding communication of diagnostic uncertainty to pediatric patients and their families.

PHM and PEM physicians had many similarities in their opinions on diagnostic uncertainty, specifically in the diagnosis and work-up for patients in whom they were unsure of the diagnosis. The groups diverged in their perspectives on the goal of the encounter. PEM physicians valued disposition over diagnosis, because their primary aim is to determine whether the patient needs to be admitted or discharged from the ED. PHM physicians had the ability to observe the progression of patients with diagnostic uncertainty or the test of time. This may speak to the differences in management styles in these 2 subspecialties.

Our study has several limitations. Our focus groups were conducted at a single institution, so these opinions might not be reflective of other pediatric physicians at different institutions.

Second, our focus groups only consisted of pediatric emergency medicine and pediatric hospital medicine physicians; nurses and other allied health professionals are part of the diagnostic team and may have different and complementary views. Third, our focus group consisted of attending physicians; trainees likely have different perspectives on diagnostic uncertainty. Fourth, the focus groups were moderated by a hospital medicine physician, which could have potentially biased the discussion. Fifth, we did not include pediatric physicians from other specialties, and we recognize diagnostic uncertainty is seen in all specialties. Finally, because our focus group participants were volunteers, nonparticipants may have different beliefs because of selection bias.

Pediatric emergency medicine and hospital medicine physicians shared common themes in approaching diagnostic uncertainty, which we mapped to different phases of a model for the diagnostic process. Management strategies employed by physicians included asking colleagues for guidance, making sure to rule out diagnoses with significant morbidity and mortality, and contingency planning. PEM physicians relied on disposition over diagnosis, whereas PHM physicians used test of time for managing patients with diagnostic uncertainty. Physicians expressed reluctance to frankly disclose diagnostic uncertainty with patients and their families; additional study of best practices in communicating diagnostic uncertainty is needed.

We are grateful for the immense scientific contributions of Dr Steven T. Chan to this manuscript.