Documentation as composing: how medical students and residents use writing to think and learn

Some educators have described clinical documentation as “scut”. Research in medicine has focused on documentation’s communicative value and not its function in learning. With time being an important commodity and electronic health records changing how we document, understanding the learning value of documentation is essential. The purpose of this study was to explore how trainee composing practices shape learning. Qualitative methods employing Rhetorical Genre Theory were used to explore clinical documentation practices among medical trainees. Data collection and analysis occurred in iterative cycles. Data included field notes and field interviews from 110 h of observing junior trainees and senior internal medicine residents participating in patient admission and follow-up visits. Analysis was focused on Paré and Smart’s framework for studying documentation as composing. From a composing lens, documentation plays a vital role in learning in clinical settings. Junior trainees were observed to be reliant on using writing to support their thinking around patient care. Before patient encounters, writing helped trainees focus on what was already known and develop a preliminary understanding of the patient’s problem(s). After encounters, writing helped trainees synthesize the data and develop an assessment and plan. Before and after the encounter, through writing, trainees also identified knowledge and data collection gaps. Our findings highlight clinical documentation as more than a communication task. Rather, the writing process itself appeared to play a pivotal role in supporting thinking. While some have proposed strategies for reducing trainee involvement, we argue that writing can be time well spent.


Introduction
In clinical learning settings, time is a precious commodity (Block et al., 2013;Chaiyachati et al., 2019). While direct patient care activities are typically viewed as time well spent for both learning and patient care, indirect care activities such as clinical documentation have task expectations are and; how to prepare for the collaborative aspects of composing (i.e., case review).
To further our understanding of the learning value of writing during clinical work and to inform education practices around clinical documentation, the purpose of this study was to explore trainee composing practices in the context of a clinical setting where composing is used to perform clinical work.

Method
This was a rhetorical genre study based on the research methodology proposed by Paré and Smart for studying composing, which is a theory-informed method for observing and exploring genre composition in real-world settings (Paré & Smart, 1994). The research team consisted of a pre-clinical medical student (DB) who was trained to participate in observational fieldwork by the senior author (MG)-an experienced qualitative researcher with expertise in both medicine and RGT-and a research associate (JT) who worked with MG and who is also experienced in qualitative observational research.

Setting & sampling
Between June 2018 and August 2019, data were collected from six internal medicine inpatient teaching teams spread across two hospitals of an academic health sciences centre in Ontario, Canada.
Both purposeful and theoretical sampling were employed. Purposeful sampling was used at the trainee, location and shift levels. Trainees were sampled to capture a range of experience across and within years. Sampling from two different hospitals accounted for cultural and workplace practice variations, while observation of both daytime shifts and overnight call captured variations in workflow and team interactions. Later in data collection, we also engaged in theoretical sampling to help enrich and refine our emerging understanding of the phenomenon. For example, over time we recognized the need to focus more heavily on observations of on call shifts instead of daytime shifts. We were also able to incorporate questions into our field interviews to better interrogate aspects of our data that we had observed in previously but did not fully understand or to test our evolving theories. This iterative process thus allowed our data to adapt and properly inform our emerging understanding.
To minimize participant reactivity (Paradis & Sutkin, 2017), we conducted sustained observations and limited participants' awareness of our specific research goal to further mitigate this concern. In obtaining consent, participants were informed about a more generic study goal of observing communication and documentation practices.

Data collection
A medical student collected data as they were able to build rapport with trainees during extended observations, understand the medical context, and create a non-evaluative environment. Consistent with RGT informed research principles (Charmaz, 2014;Schryer, 2011), our data were collected through fieldnotes from direct observations and audiorecorded field interviews on the inpatient wards and in the emergency room.

3
A total of 26 observations were conducted. We observed six senior medical students (clinical clerks (CC)) rotating through their internal medicine core rotation, six junior internal medicine residents in their first post-graduate year (PGY1), one family medicine resident (PGY1) and eight senior internal medicine residents (3 PGY2s, 5 PGY3s). Eight field observations were conducted with junior trainees (medical students and PGY 1 residents) during daytime shifts on the wards. A further 18 observations were conducted during overnight call (junior trainees) and Emergency Department Consultation (senior trainees) shifts. During daytime shifts, trainees were observed during morning rounds, then shadowed as they saw assigned patients. During on-call shifts, trainees were observed at handover, then shadowed for several hours as they consulted on new admissions and managed overnight inpatient issues.
In both settings, the focus of observations was on the trainees' multiple stages of composing. Observations occurred over more than 110 h, with one to four shifts observed for each participant. Documentation practices observed included both written notes (e.g., daily progress notes) and dictated notes (e.g., admission notes and discharge summaries). Field interviews accompanied observations -brief interviews of 1-10 min in duration performed, as time allowed, and at multiple time points throughout any given observation period. Using a live scribe pen and notebook ("8 GB Echo Smartpen-Livescribe Smartpen-Livescribe Inc. (US)," n.d.) so as to make recording unobtrusive, these interviews, when possible and with prior consent, were audio-recorded and subsequently transcribed by DB. Through these field interviews, the thinking behind practices was made visible. This approach is very similar to the one used by clinical reasoning researchers (Koufidis et al., 2020;Pinnock et al., 2019). Early in the process, the field interviews were used to explore observational findings. Over time, with repeated exposure and analysis of existing data, thinking could be largely inferred based on observations of how a particular task was carried out. For example, when trainees are observed to go back and forth between the medication list and the past medical history, it is clear that they are trying to reconcile which medications go with each of the patient's chronic active problems. Similarly, there is a large difference between writing out the list of medications in any order versus reconfiguring the list so that medications for each chronic active problem are kept together.

Analysis
Principles of rigour, similar to those posited by constructivist grounded theory, were used to inform data collection and analysis, which took place in iterative cycles with the analysis of existing data informing subsequent data collection (Charmaz, 2014). Reflexive memos were used to maintain an understanding of the research team's position relative to the participants. For example, DB, being a medical student, used memoing to grapple with his observations and ensuing judgements. Early on, these memos and reflections were used during team meetings to unpack some of the complexity of clinical practice that he was observing. This allowed him to uncouple the clinical practice and his own knowledge gaps from the documentation practices themselves, which were the focus of the study. We also employed constant comparison, analyzing earlier data for new themes that emerged later in the process.
For the initial analysis, thematic coding of three field notes was conducted independently by DB to generate a broad set of codes representing phenomena observed in the field in relation to composing practices. 16 The initial coding was conducted manually, going line by line through the transcripts. These codes were then discussed with MG and 1 3 JT. The initial set of codes was condensed and then focused coding was applied by DB to additional transcripts. A series of regular meetings-10 in total-were had between the research team to review and refine the coding process and develop an understanding of how composing is enacted by junior and senior trainees on the internal medicine service. Through the ongoing cycles of collection and analysis, the codes were refined and used to develop a theory of how trainees composing relates to learning. Theoretical sampling was also used to inform further sampling and identify and reconcile discrepant data.

Results
Across the observations, we were able to identify, confirm and contextualize all of the composing processes described by Paré and Smart. The processes were identified across the data, including admission note writing by junior and senior trainees, ward progress notes written by junior trainees and ward notes written when junior trainees were called to assess a patient. For junior trainees, composing always included writing, as will be elaborated on below. For routine cases, senior residents could compose by assembling documents from the medical record and dictating from these without first creating a written template.
Regardless of training level, trainees appeared to use and described using composing to help frame their thinking about a patient. Depending on the individual trainee, however, more or less time appeared to be spent thinking; for some of the junior trainees, the task was approached in a more formulaic way, trying to add something, in writing, into every section of the note but without necessarily attempting to make sense of it all.
Two important periods were identified and self-evidently labelled before and after the patient encounter. Both periods contained a set of tasks that trainees used writing to help them address. Some were unique to one of the periods, while others were present in both. Additionally, we identified differences in how they approached tasks in the time periods before and after the patient encounter and how this changed with experience. The following sections will explore the tasks seen in both periods more thoroughly. Short representative quotations are used throughout to illustrate key findings. In these, we use the short form CC (clinical clerk) to represent medical students, IM to represent internal medicine residents and PGY1-3 to represent the post-graduate year of the resident being observed. The numbers after each abbreviation represent their random number generated ID. For those interested in more detailed data, the more elaborated quotations have been included in an online Appendix Tables 1, 2 and 3. The table, section and row where they are found are indicated at the end of the quotation (e.g., Table 1-1A).

Before the patient encounter
Before all encounter types, composing was focused on understanding what is already known about the patient and guiding what needed to be collected. In new cases, this related to past medical history, medications and available results of investigations. By contrast, what was reviewed during follow-up cases on the wards depended in part on how well the patient was already known to the trainee. Thinking tasks that were observable included forming a preliminary understanding of the patient's contextual features (e.g., living situation, social supports, etc.), making connections between the patient's medications and their past medical history, developing an understanding of the patient's problem(s), and flagging gaps that exist within their own knowledge.
By combining knowledge of the presenting complaint, or known active issues, with the context provided by older notes, trainees could also, in part, plan what they needed to determine during their encounters. Junior trainees commonly wrote down notes on pertinent information collected before the patient encounter. They would then use this to guide what still needed to be collected or verified during the encounter. During new cases, for example: My preference is to write things out beforehand so that I have a thorough understanding of the patient's backstory as much as possible and then I can complete details and verify details… just for efficiency's sake as well as continuity's sake, and safety's sake, I write it out first, verify with the patient and then fill in details as needed. (Table 1-1B) Field interview 13, IM576 (PGY1). Trainees across all levels employed a similar approach, but there were noticeable differences across levels of experience. Less experienced trainees were less selective in their collection and analysis of information, often electing to read and record all available information: He then opens a discharge summary from a few years back. Noting the comments on Chronic Obstructive Pulmonary Disorder (COPD), he comments that he needs to read more... writes TIA [transient ischemic attack] on his chart paper, below COPD. He then adds the ulcer below TIA, re-ordering the issues from the discharge summary… (Table 1-1C) Field visit 5, CC977 (Medical student). Trainees with more experience also reviewed pre-existing notes. However, they were more selective in what they considered, trusted and wrote down: If it's [medication list] more than a month old, I feel like things could have changed. And even then, I don't try to rely on it too much, because there can always be errors in dictation… (Table 1-1E) Field interview 26, IM738 (PGY2). In addition to accurately capturing the information, reviewing and writing out the past medical history and medication list for a new patient, especially for more junior trainees, could be used to help them to link the medications with chronic active problems: Field interview 23, IM133 (PGY2). The review and writing out of the medications could also be used to identify active issues or potential causes for active issues: He reads over the list of meds on the summary and begins to write them out… He gets seven of the nine down and then stops to wonder aloud, "Why is she on [this medication]?" He then flips over to a service consult note and begins to read through it… (Table 1-2A) Field visit 9, IM939 (PGY1).
The pre-review of existing clinical investigations could be used to help identify active issues or potential causes of active issues. It could also be used to guide early ordering of new investigations in the emergency department or on the wards: …She highlights one line and appears to read it more closely. She then grabs the phone and calls the core lab, requesting that they add additional bloodwork for the sample they have drawn from this patient. (Table 3-2C) Field visit 22, IM289 (PGY3) … She pauses as she sees that the urine electrolytes and osmolality have not been checked yet and comments that these need to be done. She adds an order for the required tests. (Table 3-2D) Field visit 24, IM425 (PGY2). While the most experienced trainees that we observed wrote less down during prereview, they were still observed to spend considerable time reviewing patient records before encounters: He spends about 2 minutes reading this note before exiting and opening an older progress note... Throughout this entire process, he has not written anything down…. (Table 1-1F) Field visit 18, IM865 (PGY3).

After the patient encounter
After talking to and examining the patient, trainees shifted their focus to a different set of tasks. While trainees now needed to document a complete note with all of its traditional sections, the primary focus was on making sense of-or for follow-up cases updating-the problem list and formulating an assessment and plan. After the patient encounter, synthesizing the data and formulating a problem list was a key focus. The first step in this process was organizing all patient information, including details from the history, physical exam and investigations. For junior trainees, writing was consistently used to assist with this, at times, lengthy and thinking-intensive process: Sitting down at the computer station, CC977 prepares to work through the information he has collected… He begins to reflect aloud on what he wrote, verbally walking through some of the findings… After reflecting on it, he explains to me that he likes to talk it over to see if he has it all … Upon reflection, we return to the room to ask another question and CC977 immediately adds the patient's answer into the chart. We leave the room and CC977 begins to write down some more exam findings in the appropriate section of his note. (Table 2-1A) Field visit 5, CC977 (Medical student). With the most junior trainees, documentation sometimes proceeded in a more formulaic way, with a focus on writing to fill in the expected values of a section before making sense of what the written information itself meant: She then begins to add bloodwork values, using the computer that is still logged into Power-chart... She then jumps back to the A/P section and adds M [for metabolic], writing down notes on the presence of an electrolyte disturbance. She pauses and then adds a note on the need to replace the electrolytes. She then paused for a moment to look at the electrolyte values and turned to ask the SMR about what she was seeing. (Table 2-1B) Field visit 9, CC100 (Medical Student). Synthesis appeared to evolve with increasing experience. More senior residents were less tied to writing all details prior to synthesizing. In part, this was an evolving necessity as Senior Medical Residents [SMR] acknowledged needing to be more efficient given an increased workload: So as a PGY1 I used to do that. I would pretty much have a fully handwritten note. But now that I'm PGY2, especially as the SMR, it's too busy. I can't take the time to write something down... I feel pretty comfortable… I know the kind of checkboxes that I have to hit… (Table 2-1C) Field interview 26, IM738 (PGY2). The development of an assessment and plan tended to follow the synthesis process described above and involved taking time for reflection:

He then adds the heading for assessment and plan. He writes a summary of the patient across the top of this section… He adds 1) Weakness, then pauses and comments he is not sure what the cause is. The patient had described what she thought was the problem [Coumadin prescription] but he cannot make himself believe that's it. He then returns to his note and adds two queries after the issue [similar process for 5 more issues]. He skims through what he wrote. Turning to the computer, he opens the labs page and reads through what has been done. He
debates out loud whether he should do more bloodwork. He then flips over to the orders page and looks at the pending queue and makes note of some of the tests related to earlier issues. He turns back to his page and reads it over. (Table 2-2A) Field visit 10, IM939 (PGY1). Regardless of context-follow-up on the wards or new patient in the emergency department-for junior trainees, the effort expended during composing appeared to influence the ensuing teaching and guidance received around a case. Those who invested less time and effort developing their ideas before presenting and those who felt pressure (usually internal vs. external from the SMR) to prepare quickly appeared to have less meaningful dialogue; instead, they would frequently be told what to think or do: As she writes out this issue, CC100 is just writing down what the SMR is telling her... (Table 2-2B) Field note 9, CC100 (Medical student). For a new patient, in addition to case review, another observed activity that reflected the effectiveness of early composing practices was dictation. For medical students and some junior residents, their approach to dictation was to write notes by hand and then read their notes as a script during dictation. This was seen as advantageous because it helped organize their thoughts and ensure they didn't miss pertinent points: CC681 reels off the identification information very quickly, with no issues or pauses as he talks. He then begins to read over his notes as he continues into the dictation. (Table 2-2C) Field visit 17, CC681 (Medical student).
Over time, however, residents began to reduce the level of detail in their handwritten pre-dictation admission notes. Instead, they used them to map out the pertinent features of a case and organize their thoughts: Because you get it within two hours of dictating your note, it's mostly just for myself to organize my thoughts... So, I write down the pertinent negatives… But other things that are less pertinent or other negatives you just go through, usually just kind of review of systems things, I don't really write them… I write down the most important negatives and then the positives… (Table 2-2E) Field interview 10, IM939 (PGY1). Throughout residency, trainees described a gradual progression in which they became able to handle thinking through their cases with less reliance on writing. This was true for routine cases and when there was readily available information in the electronic health record (

Identifying gaps
Before and after the patient encounter, trainees also identified gaps in personal knowledge and data collection. Examples of such gaps might include not recognizing a medication or being unfamiliar with a particular medical problem:  (Table 3-1B) Field interview 5, CC977 (Medical student). In the case of data collection gaps, there were multiple points in the composing process which facilitated the recognition of these; most junior trainees were frequently observed to recall additional pieces of information that were required while writing a clinical note: As she is reading the chart and documenting her findings, she realizes that she has not done a full neurological exam and will need to go back to do one on the patient. (Table 3-2A) Field visit 9, CC100 (Medical student).
Data collection gaps for junior trainees were also commonly noted during the process of case review: Scanning through the chart, IM605 interrupts the presentation to comment that there is a note in the chart about a heart condition. CC594 does not appear to have previously read about this condition… IM605 then tells CC594 to go back and see the patient again, to clarify a few of the things the presentation has shown he did not obtain the first time. (Table 3-2B) Field visit 14, IM605 (PGY3), CC594 (Medical student).

Discussion
To broaden the debate on the educational value of documentation, building on work from the field of rhetorical genre studies, we explored the relationship between writing and thinking in junior and senior medical trainees. In addition to confirming and contextualizing the Paré and Smart model of composing practices (Paré & Smart, 1994), our study demonstrates the important and complicated developmental relationship between writing and thinking in the clinical setting. As some have problematized the educational value of documentation and the proportion of time it occupies, (Dresselhaus et al., 1998;Petrany, 2013) our findings also have implications for three inter-related discussions in the field: scut work, medical scribes in residency training, and electronic documentation.
Our key study finding is that for clinical trainees and educators, the term clinical documentation may represent a problematic misnomer serving to direct our attention to only one of its purposes-documentation requirements. We would argue that the better term is composing, which captures the more complex interaction observed between writing and thinking. While the most senior residents consistently appeared to be able to make do with more minimalist writing during the composing process, this was only possible for patients where existing clinical notes on the electronic health record provided most of the required contextual data (i.e., past medical history and medication list) and, when the patient cases represented relatively straightforward problems from that resident's perspective. For all other trainees, writing played an essential role in supporting their thinking in their patient encounters.
Our findings also confirm Paré and Smart's model of composition and draw attention to several contextual features that may help the struggling trainee and those new to the clinical environment. First, information gathering and analysis are an iterative process that begins before seeing the patient and continues through to after the initial encounter is complete-some trainees need to return to the bedside more than once to gather and confirm data. Second, junior trainees require considerable time post-encounter to write and think as they make sense of the data. Finally, shortchanging the thinking process has consequences; whether it occurs through early prompted review or through self-initiated early review, skipping the key thinking steps led to very different case reviews; ones characterized by minimal discussion and teaching and more telling of what to write and what to do.
While many of the observed junior trainees used a focused approach-an approach guided by the patient's identified issues-to guide their writing and thinking prior to seeing a patient and following an encounter, not all trainees recognized or meaningfully engaged in the associated thinking tasks. Rather, similar to findings from Cadieux et al. (Cadieux & Goldszmidt, 2017), some medical students and junior residents appeared to not recognize the associated thinking required and instead approached composing as a formulaic documentation task predicated on filling in information into key sections of the note.
Both our findings and prior research suggest that learning composing occurs through trial and error, with inconsistent effort to directly tell trainees where their priorities should be and why (DeLeon et al., 2018;Rowlands et al., 2016). While prior efforts have shown that documentation quality can be improved with feedback (DeLeon et al., 2018;Opila, 1997;Rowlands et al., 2016), we would suggest that using our findings to teach trainees about composing might help them to not only improve documentation quality but also to better understand where they should focus their attention.
From an RGT perspective, there is an important and complicated developmental relationship between composing, thinking and attention that argues for the importance of composing beyond its direct influence that we studied (Bawarshi, 2001;Bazerman, 2009). During composing, according to Charles Bazerman, a leading RGT theorist: "genres identify a problem space for the developing writer to work in [e.g., focusing attention on the tasks associated with admitting a patient to hospital] as well as provide the form of the solution the writer seeks and particular tools useful in the solution. Taking up the challenge of a genre casts you into the problem space and the typified structures and practices of the genre provide the means of solution [e.g., following the admission genre can act as a scaffold to support thinking] (page 291)" (Bazerman, 2009). While not explicitly addressed in our study, according to RGT, composing's developmental influence extends beyond the task itself. Over time, composing with the genres of the profession supports trainee socialization and identity development: "When communicants use genres, they are interpreting and enacting the social motives (embedded rhetorically within it) that sustain an environment and make it meaningful, and so are becoming socialized into producing not only certain kinds of texts, but also certain kinds of contexts, practices, and identities-ways of being and acting in the world, socially and rhetorically (page 78)" (Bawarshi, 2001). Therefore, the more a trainee grapples with the development of expertise in the profession and how to effectively communicate it through the genres of the profession, the more they begin to think like and enact the roles of the profession.
Some have considered documentation to be a "scut" like task in medicine, with little educational value relative to the proportion of time it occupies (Dresselhaus et al., 1998;Petrany, 2013). This perspective has also led some to suggest the use of medical scribes to reduce trainee time spent on "low yield" documentation activities (van Schaik, Reeves, and Headrick 2019). Our findings argue against this stance, and we would caution against the use of medical scribes in all but the most advanced trainees. While it is true that, when done well, writing can be time-consuming, it is necessary developmental work and should not be dismissed. Over time, senior trainees appeared to develop the ability to abbreviate the written component of composing. However, these trainees commented that writing was essential in their personal development and that they could not perform at their current levels of proficiency without first learning how to do it through writing.
While our study did not take place in a context where trainees entered text directly into the electronic health record-they dictated their clinical notes-our findings may have important implications for contexts where they do. Others have written about concerns related to cut and paste and the propagation of errors (Koppel, 2014;Siegler & Adelman, 2009), the use of templates frequently conveying false negatives (J. E. Siegler, Patel, and Dine 2015), and templates and checkbox menus interfering with trainees independently learning which items they need to document (Mintz et al., 2009). What our findings add are concerns related to cut and paste or pre-populated fields and the ways they might interfere with the necessary thinking that junior trainees may not recognize that they need when 1 3 seeing patients. While some institutions have policies prohibiting medical students from copying information due to the risk of inaccuracies in data collection (Weis et al., 2014), we would argue that an equal concern is its impacts on learning and the provision of care.
It is important to consider our findings within the limitations of our study. Our results were obtained exclusively from Internal Medicine and it is unclear how these findings may transfer to other specialties. Additionally, while we can speculate on implications for contexts where direct electronic health record documentation takes place, unknown are the possible alternative strategies that trainees might develop for engaging in the necessary thinking when writing certain sections of the document is unnecessary. Finally, our study took place in a context where documentation was very much directed towards patient care. However, we recognize that regulatory and billing requirements in some countries may lead to trainees being involved in other forms of documentation that may not be as clearly linked to clinical care. In these contexts, it may very well be appropriate to consider strategies for reducing trainee involvement in certain forms of documentation to enhance the time they can spend on more meaningful activities like composing.

Conclusion
Our findings demonstrate that the processes of writing and thinking are tightly linked in medical trainees through the process of composing and that the process of writing during composing is an essential developmental step for junior trainees. While some have described documentation as a 'scut' like activity and have proposed strategies for reducing trainee involvement in it, we would argue that writing is thinking and learning and that it can be time well spent.

Declarations
Competing interests The authors have no relevant financial or non-financial interests to disclose.
Ethical approval This study was reviewed and approved by the University of Western Ontario Health Sciences Research Ethics Board (Project ID 6822).
Informed consent Informed consent was obtained from all participating individuals in this study. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.