The Keyword Effect: A Grounded Theory Study Exploring the Role of Keywords in Clinical Communication

Abstract

Objectives

Keywords, also known as “buzzwords” or “trigger words,” serve as memorable descriptors to associate physical findings with specific diagnoses. These terms, such as “target lesion” and “steeple sign,” liken a physical or radiographic finding to a nonmedical comparator as a means to elicit an associated diagnosis. Keywords permeate medical literature and clinical conversations. However, the potential for miscommunicating critical information exists and the impact of keywords on communication‐related medical error is unknown. We explored the use of keywords and how physicians perceive their use in their clinical communication as part of patient care.

Methods

With a grounded theory approach, semistructured interviews were completed in 2016 to 2017 with a purposive sample of 15 resident and attending physicians working in one pediatric emergency department, where clinical conversations occur frequently between providers of different specialties and levels of training. Constant comparative analysis for emergent themes was conducted. We identified key themes and examined their relationships to theorize how keywords affect clinical communication.

Results

We identified three major aspects: 1) keywords belong to the culture of medicine, by which providers connect with each other using specialized terms that imply a shared experience and knowledge base. This culture encourages keyword use. 2) By encapsulating a pattern of clinical findings into one word or short phrases, keywords allow for convenient, efficient communication of both diagnoses and of thought processes between providers. 3) Keywords, however, may mislead; if incorrectly applied to a given clinical situation, they may be misinterpreted by the receiver, or they may introduce bias to diagnostic decision making.

Conclusions

More than simple descriptors, keywords can communicate entire diagnoses and activate illness scripts between providers. Also, keywords are integral to the culture and language of medicine. However, providers should be aware of the potential negative effects of keywords in clinical conversations and must balance the demands of efficient and accurate communication with the potential for miscommunication and error.

Keywords, also known as “buzzwords” or “trigger words,” are terms used as memorable descriptors to associate physical findings with specific diagnoses. These descriptors, such as “butterfly rash,” “honey‐colored crust,” and “hot‐potato voice,” liken a physical or radiographic finding to a nonmedical comparator and are closely associated with a single diagnosis (Table 1). While keywords permeate medical literature and daily clinical conversations, these terms may lead to medical errors due to their associative properties, contributing to cognitive bias or misdiagnoses based on false assumptions. ¹ , ² , ³ In addition, keywords may contribute to miscommunication, although the impact of keywords on communication‐related medical error is unknown.

Table 1.

Examples of Keywords Used in Clinical Communication

Keyword	Associated Diagnosis
“Butterfly” rash	Systemic lupus erythematosis
“Christmas‐tree” pattern rash	Pityriasis rosea
“Currant jelly” stool	Intussusception
“Honey‐colored” crust	Impetigo
“Hot potato” voice	Epiglottitis
“Steeple sign” on neck radiograph	Croup
“Target” lesions	Erythema multiforme

Keywords appear widely, including published review books, textbooks, and journal articles. ⁴ , ⁵ , ⁶ They are often introduced as tools to assist in recalling diagnoses for knowledge‐based examinations. ⁶ Multiple authors explored the use of keywords as a mnemonic device to improve recall. ⁷ , ⁸ , ⁹ However, in clinical medicine, these terms are not limited to diagnostic tests; they permeate medical conversations between providers as a communicative tool. The concept of keywords and their potential cognitive effects appear in educational literature. For example, Chang et al. ¹⁰ deliberately avoided keywords when writing clinical vignettes, as encountering keywords “negated the need for the entire vignette.” Yet, the specific role that keywords play in communicating information and how they affect the receiver’s perception of that information has not been explored directly.

While keywords enhance recall, we need to understand better the risk of miscommunicating critical information when using keywords as part of patient care. Communication failures can lead to medical error, ¹¹ whether due to interprofessional and interpersonal dynamics ¹¹ , ¹² or by misinterpretation of medical language. ¹³ As keywords feature prominently in published educational materials, ³ greater insight into how keywords affect communication would help clinical educators approach keyword use in health care settings. Thus, we aimed to explore how attending physicians and residents use and perceive keywords in their clinical communication.

Methods

We used a constructivist grounded theory (CGT) approach, a qualitative methodology suited to explore and explain social phenomena. ¹⁴ Because CGT relies on interpretive techniques that reflect the perspectives of both the subjects and the researchers, we provide the following information: both authors (MC and WE) are pediatric emergency physicians whose clinical work involves patient care and the supervision and education of medical trainees; both have experience in educational research, and WE has extensive experience with qualitative methodology.

For the study setting, we chose the pediatric emergency department (ED) at Ann & Robert H. Lurie Children’s Hospital of Chicago, a teaching hospital of Northwestern University for several reasons. First, we located our study in this setting based on the high frequency of conversations involving the communication of clinical information between trainees and clinical supervisors. For example, resident physicians typically assess patients before presenting key information to attending physicians. Second, residents often seek additional guidance or management expertise from subspecialists by telephone or in person. Finally, the high patient turnover in the emergency setting further increased the opportunities for patient encounters in relatively short time frames. An initially purposive sample of residents and faculty was recruited from the departments of pediatrics and emergency medicine at Northwestern University Feinberg School of Medicine. We deliberately selected subjects to gather perspectives from physicians working in the same clinical setting but with differences in training background and level of experience. There was a sizeable pool for recruitment of faculty physicians and both pediatric and emergency medicine residents from this department. We recruited residents from both pediatrics and emergency medicine training programs. All attendings working in this department were board‐certified in pediatrics or board‐certified in both pediatrics and pediatric emergency medicine. In later stages of data collection, we used principles of theoretical sampling in line with CGT to guide additional subject recruitment.

Subjects were recruited via an e‐mail that introduced the study and invited their participation. Each participant was assured of confidentiality and anonymity. Informed consent was obtained prior to each interview. The study was granted exempt status by the institutional review boards of Northwestern University and Lurie Children’s Hospital.

We used semistructured interviews to allow for flexibility to explore emergent themes (see Data Supplement S1 [available as supporting information in the online version of this paper, which is available at http://onlinelibrary.wiley.com/doi/10.1002/aet2.10424/full] for interview guide). Clinical vignettes were used to establish and verify familiarity with the principle of keywords, and then open‐ended questions invited participants to share their experiences and reflections on keyword use. The keywords used in the vignettes were selected by a focus group of pediatric faculty from a list of published keywords. ⁶ The interview guide was developed to elicit perceptions related to various facets of keyword use and was developed iteratively in multiple test phases with subjects who not included in the final analysis. Due to iterative data collection and analysis, the interview guide evolved throughout the study based on our evolving conceptual understanding of keyword use.

Consistent with CGT methodology, data collection and analysis occurred iteratively, with data analysis informing our subsequent approach to data collection. Interviews were performed by a single investigator (MC) and were audio‐recorded, transcribed, and anonymized. Both investigators read and analyzed initial transcripts independently, meeting regularly to compare analyses. Subsequent analysis was conducted in phases, starting with line‐by‐line coding (by MC) and using constant comparison to develop focused codes. These codes were subsequently coalesced into larger thematic categories (MC and WE), which were further refined into a conceptual model (MC and WE). ¹⁴ , ¹⁵ Disagreements in analysis were resolved by discussion. Data collection continued until theoretical sufficiency was reached ¹⁵ and no new insights or themes were obtained from interviews.

Results

Fifteen subjects (seven faculty, five pediatric residents, three emergency medicine residents) participated in individual, semistructured interviews lasting up to 45 minutes. Two subjects, both faculty, completed their initial interviews and were invited to follow‐up interviews to explore additional emergent themes and insights discovered during subsequent interviews with other faculty and residents.

We identified three major themes (Figure 1):

1. Keywords are embedded in the culture of medicine.

2. Keywords allow for efficient communication.

3. Keywords can be misleading.

Figure 1.

Theoretical model for the effect of keywords on clinical communication.

We will now discuss these themes in greater depth and include representative quotations from our interviews to illustrate our main findings.

Keywords Are Embedded in the Culture of Medicine

All participants expressed familiarity with keywords, regardless of specialty or level of experience. Physicians described a variety of settings in which they had either seen or used keywords, including studying for tests, teaching others, documentation in the medical record, and verbal communication about patient care. One provider enthused,

I use it in charts. I use it to describe findings. I use it during signout. I use it in teaching. I don’t try to avoid them. (Attending 5)

Statements such as this clearly established the ubiquity of keywords within medical language, demonstrating their usefulness and convenience. Further, residents and attending physicians both linked keywords to studying for examinations during medical school, particularly during the preclinical years:

I feel like a lot of medical learning, especially medical school, is associated with buzzwords for exams or tests. (EM resident 2)

Using keywords to quickly recall specific diagnoses was a strategy that was emphasized in published board‐study materials. The power of keywords to associate findings with specific diagnoses was therefore seen as a useful tool in teaching, particularly in preclinical years. One resident recalled:

In medical school, you do so much studying with books and papers, and what is that, Step 1 …, in the back is literally a page of coined phrase associations. (pediatric resident 5)

Supervising physicians reported using learners’ familiarity with keywords to their advantage to teach and connect with students. One attending physician explained,

I learned them, they [students] learn them, they love them. They’re like “oh I can say that! I got it right!” It’s a great little pearl to share that you both can bond over. (attending 6, interview 2)

The “bond” that this physician describes exemplified the importance of keywords within the culture of medicine. In addition to communicating information, keywords represented a specialized, colloquial vocabulary exclusive to clinical providers. According to one attending: “it’s a part of our language” (attending 6, interview 1). Another participant explained, “it’s important to learn how to communicate what we see using a universal medical language” (pediatric resident 5). By using keywords, providers connected with each other through words that implied a shared experience and knowledge base.

Keywords Allow for Efficient Communication

The associations between keywords and diagnoses allow for convenient, efficient communication in appropriate cases. Here, “Hear X, think Y” described not only the connection between the finding and its diagnosis but also the automatic and quick mental association that occurs upon hearing the keyword. This association seemed instantaneous and effortless. For example, when discussing the connection between the term “currant jelly stool” and the diagnosis of intussusception, one resident explained,

When I hear that word, it makes me think of that [diagnosis]. Hear X, think Y … I think currant jelly stool is one of the big ones that comes to mind. It’s a clear association. (pediatric resident 3)

Keywords, therefore, helped communicate more than just a diagnosis. By instead encapsulating a thought process, they were used to build a differential or justify a plan of action in a single short phrase. In this light, keyword use was viewed as advantageous by quickly and efficiently communicating information between colleagues. For example, one subject explained,

I think it is helpful to be able to say a word and for someone to know exactly what you’re thinking about right away, with a word. (pediatric resident 4)

Further, residents expressed the importance of demonstrating to their colleagues and supervisors not only what they thought was happening, but why. One resident explained:

In signout, when we justify what’s happening with a patient, or we rationalize our medical decision‐making, a lot of times we’ll throw in words or say things that are not happening, to let other providers know why we’re doing what we’re doing. “This person is not ‘toxic‐appearing.’” “They don’t have a ‘hot‐potato voice.’” We use a lot of buzzwords that we are all familiar with to help other practitioners see why we’re doing the things we’re doing or not doing the things we’re not. (EM resident 1)

Keywords Can Mislead

Despite their utility in communicating efficiently, keywords can also be misleading in certain contexts. Specific keywords may be nonspecific, and in some cases can lead to anchoring bias. Providers offered several instances in which they avoided using keywords due to the potential for miscommunication. Further, keywords were also misapplied, misinterpreted, or contributed to anchoring bias and premature closure. One attending reported:

“It may not be communicating what you think it’s communicating.” (attending 3)

Incorrect Application

Some providers expressed concern that keywords may be used for the wrong trigger. In other words, a clinician may mistakenly use a keyword to describe a finding, leading others to assign an incorrect diagnosis based on the keyword. The potential for assigning the incorrect keyword seemed to be greater with keywords describing rare findings or with providers with less clinical experience. One attending physician warned,

You have to be careful of [keyword] use as a student and make sure that what they’re using a keyword to describe is actually, truly associated with what their findings are. Because if they don’t understand what their findings are, then they can use it inappropriately. (attending 5)

Not all keywords were equivalent in their associative utility, particularly if the comparator was itself an unfamiliar object. According to our subjects, providers, and particularly learners, should be mindful to use keywords appropriately when describing findings. For example, one resident explained that his own experience with currant jelly might color his interpretation of the phrase “currant jelly stool”:

How many people have actually seen currant jelly stool? … Am I even going to recognize currant jelly if I see it? I’m not quite sure. (EM resident 1)

Conversely, another resident expressed greater comfort with the phrase “soap bubble appearance” when describing the radiographic appearance of pneumatosis intestinalis on abdominal radiograph:

I can relate to real soap bubbles … And I can relate to what that would look like in the intestinal wall, since I know what real soap bubbles look like. I think that was my problem with currant jelly—is that I don’t know what currant jelly is. (pediatric resident 3)

Incorrect Interpretation

Some residents admitted that they could misremember the correct associations between keywords and their diagnoses. One mentioned,

Even strawberry tongue, I can’t remember which one it’s supposed to be associated with. So I feel like there’s going to be other situations where people are like “oh I remember, currant jelly is associated with … which one was it?” and then we’ve lost it. Instead of saying “we should be suspicious of intussusception when we see stool that’s concerning for blood.” (pediatric resident 3)

Participants also suggested that providers of different subspecialties may lack equal familiarity with keywords or may have dissimilar keyword vocabularies based on their unique training and experience. One resident reported:

If they’re from a specialty that doesn’t use that word, or I don’t think they use that word, or if they’re from a training area or part of medicine where they aren’t exposed to the same training as me, I’ll be less likely to use it. (EM resident 1)

In light of these insights, our subjects recommended considering the perspective of the person with whom they are using a keyword. Also, keywords relied on an associative link, but there may be cases where this link may not be as strong or as specific as the speaker assumes.

Anchoring Bias and Premature Closure

Our analysis suggests that providers should be careful and purposeful when deciding to use keywords. When describing physical findings in instances of diagnostic uncertainty, physicians in our study deliberately avoided keywords to help prevent others from focusing on a specific diagnosis at the exclusion of other possibilities. For example, residents and attending physicians alike warned of the potential for keyword use to contribute to anchoring bias. As one resident described an example of premature closure:

When I hear the word currant‐jelly stool, I don’t think of anything else [except for intussusception]. But when I hear blood in stool, or bloody poops, or hemoccult‐positive stool, my differential is big. (pediatric resident 5)

Here, the use of a keyword inadvertently narrows down the diagnostic possibilities being considered, possibly contributing to a missed diagnosis. Another resident mentioned, discussing currant‐jelly stool, a case of anchoring bias:

I think the issue with buzzwords is that if they’re used inappropriately or out of context, they can really bias the treatment team into treating someone down a certain path. I think one of the most dangerous things in medicine is early anchoring on something … it increases your tendency to totally wipe of the board other potential considerations that can even be more life‐threatening. (EM resident 2)

An attending seconded the warning:

You have to be very confident in your diagnosis to throw those [keywords] out, because it will make someone have a skewed perception. (attending 6)

In cases where the diagnosis was not yet confirmed, some residents reported that they would avoid using keywords at all, fearing that the use of that term might lead to an inappropriate workup. One pediatric resident linked keyword use with diagnostic confidence:

I think I would use it [the keyword] once we’ve proved it was intussusception. Once the ultrasound showed me intussusception, then when signing out the patient I’d say “yeah, and they had currant‐jelly stool.” (pediatric resident 5)

An emergency resident echoed this strategy:

I think in certain situations, they’re okay to use, but to be honest, they are things that I just try to use sparingly, and really only in the actual clinical context when I really think, after a thorough evaluation, that the patient has this condition that’s associated with that particular description. (EM resident 2)

Interestingly, in these examples, a keyword is only used retrospectively, once the diagnosis is established. Although the finding may be no different, the decision to use a keyword hinges on the ultimate diagnosis, as if knowing the diagnosis makes the keyword’s use permissible.

Providers also reported that their approach to communication through keywords changed with their amount of clinical exposure. At first, keywords are taught in medical school to aid with tests:

I think in medicine, we work on recognizing patterns. But especially when we’re first learning clinical medicine, a lot of the patterns that we recognize are in specific phrases. (pediatric resident 5)

Yet, with further training, the approach to keywords changed:

Then you go out into practice and you realize that everything’s not as clearly black or white as it is in a test question. I think that starts to alter the way that you use buzzwords or sort of influences the way that you integrate them into your practice… I think as you become a practicing physician, you realize that everything isn’t black and white, and it’s really a scale of gray. You really start to change the way you look at those things. (EM resident 2)

Providers seemed less likely to use keywords when they were unsure of a diagnosis or if the clinical situation was complex. As learners progressed in their training and gained experience, a greater appreciation of these “gray” areas altered a provider’s willingness to use keywords to communicate a finding.

Discussion

In this study, we explored of the impact of keywords on clinical communication. All subjects reported familiarity with keywords, regardless of specialty or level of training, which speaks to integral nature of keywords in clinical practice. Keywords emerged as more than memory aids; keywords represent an indelible part of medical culture that both helps and hinders effective communication of clinical information. Our findings extend existing literature in several key ways.

First, in line with sociocultural learning theories, ¹⁶ , ¹⁷ keywords reflect a key component of discourse in specific communities of practice, both mediating learning and representing ways of talking in those communities that newcomers must learn. Keywords imply a shared knowledge base and specialized, exclusive vocabulary expected between providers. Our findings build on recent related work that outlined how physicians‐in‐training learning through and learned from their work‐related telephone conversations between clinicians, in part through “learning the lingo” in ways that allowed them to negotiate hierarchy, preempt pushback, and manage uncertainty. ¹³ Since keywords have colloquial and informal touch, they allow clinicians to establish rapport through shared language and demonstrate competency in their community of practice. ¹⁸

Second, keywords help to streamline the communication of specific diagnoses and diagnostic thought processes. Due to their strong associative nature, keywords can encapsulate a clinical scenario in a short phrase that would otherwise require several phrases or sentences to describe. Thus, keywords may play an important role in communicating and activating illness scripts, organized knowledge structures that help guide clinical reasoning during patient encounters. ¹⁹ , ²⁰ , ²¹ These scripts are acquired and refined through education and experience, similar to the evolving role of keywords subjects described as they progressed through medical training. In this light, using a keyword does not merely communicate words or even a phrase; instead, keyword use provides a method for the speaker to help activate the listener’s corresponding illness script for a clinical scenario.

Third, keywords can also lead to miscommunication through several mechanisms. Incorrect application and interpretation of keywords are limitations shared with other mnemonic devices and have been described in keywords in other educational contexts. ²² This study also sheds light on a potential bias more specific to clinical medicine: anchoring and premature closure. ¹¹ Our subjects described situations where hearing a specific keyword caused them to intuitively exclude certain diagnoses from their differential, a clear example of premature closure. Others mentioned anchoring by name, wherein an assumed diagnosis prevented consideration of alternate diagnoses. Keywords, with their strong associative properties, represent an intuitive cognitive process. Intuitive (or type 1) processes are fast and unconscious, but are also prone to error. ¹ Fortunately, strategies exist for cognitive debiasing. ² For example, improved education and awareness about decision making and biases can help clinicians realize potential biases before an error is made. ²³ As keywords already exist in medical curricula, concurrent education about their potential for bias may help prevent medical error.

Limitations

Although we included providers from various specialties and training levels, our study only included a single institution, which may limit the transferability of our results to other contexts. In particular, all subjects were recruited from a single pediatric ED, which would not identify cultural differences in other care settings and with other patient populations. Furthermore, our subjects participated on a volunteer basis, which may have overrepresented certain points of view with regard to interest in education, communication, and clinical reasoning. All interviews were conducted by a single interviewer who worked in the same clinical setting as the subjects. We hope that the reflexivity required in CGT served to moderate this impact. Further, the interviewer’s insider status provided context and insight into the collection and interpretation of the data, which we view as an advantage.

We see the need for additional study on keywords and medical communication based on our findings. Future research could probe more deeply into the issues we identified, including the relationships between keywords, clinical decision making, and medical error. Our qualitative study solicited our subjects experiences with keywords; future work could use conversation analysis or observing clinical communications in real time and noting when and how keywords are used. Finally, our qualitative methodology was not suited to quantify differences between or within subject groups. We expect a more robust understanding of the keyword effect through study using both qualitative and quantitative methodologies in different provider populations and care settings.

Conclusions

Our study explored into the impact of keywords on clinical communication. Our findings show that keywords represent an integral element of the culture of medicine with both advantages and drawbacks. Keyword acquisition and use relates to clinical decision making and evolves with training and experience. As such, keywords may play an important role in communicating and activating illness scripts between providers. Keywords also potentially contribute to medical error by fostering premature closure and anchoring bias. Training in cognitive debiasing strategies should increase awareness of keywords and their potential to introduce bias. Providers should be aware how keywords effect their communication and consciously balance the need for communicative efficiency and the potential for error.

The authors thank the subjects for their sincere and enthusiastic participation and Dr. Jennifer Trainor and Dr. Rick McGee for their assistance with study design.

Supporting information

Data Supplement S1. Example interview guide.

AEM Education and Training 2020;4:403–410