Patient access to clinical notes in oncology: A mixed method analysis of oncologists’ attitudes and linguistic characteristics towards notes

Abstract

BACKGROUND:

Providers have expressed concern about patient access to clinical notes. There is the possibility that providers may linguistically censor notes knowing that patients have access.

PURPOSE:

Qualitative interviews and a pre- and post- linguistic analysis of the implementation of Open Notes was performed to determine whether oncologists changed the content and style of their notes.

METHODS:

Mixed methods were utilized, including 13 semi-structured interviews with oncologists and random effects modeling of over 500 clinical notes. The Linguistic Inquiry and Word Count program was used to evaluate notes for emotions, thinking styles, and social concerns.

RESULTS:

No significant differences from pre-and post-implementation of Open Notes was found. Thematic analysis revealed that oncologists were concerned that changing their notes would negatively impact multidisciplinary communication. However, oncologists acknowledged that notes could be more patient-friendly and may stimulate patient-provider communication.

CONCLUSIONS:

Although oncologists were aware that patients could have access, they felt strongly about not changing the content of notes. A comparison between pre- and post-implementation confirmed this view and found that notes did not change.

PRACTICE IMPLICATIONS:

Patient access to oncologist’s notes may serve as an opportunity to reinforce important aspects of the consultation.

1. Introduction

Clinical documentation, including provider notes containing a vital text summary of the physician–patient clinical interaction[1] is a critical aspect of patient care. Provider notes serve not only to officially document the patient’s condition, the provider’s recommendations, rationales for decision-making and the provider’s clinical judgment, but also to communicate this information to the care team[2]. Patients have always had a legal right to their medical records, including provider’s notes; however, patients rarely requested them due to high cost of printing, extensive time between request and receipt, and other barriers to retrieval[3, 4]. Guided by the Institute of Medicine’s recommendation that patients should have unfettered access to their own medical information[5], healthcare organizations are making health care information more readily accessible to patients through the use of web-based portals. Adoption of personal health records (PHR) is expected to exceed 75% by 2020[6], and most systems will enable patients to view their full record, including provider notes.

The Veterans Health Administration’s My HealtheVet portal (myhealth.va.gov) and the Open Notes initiative (opennotes.org) have pioneered the effort to provide patients with easy access to their provider’s notes. Both initiatives encourage health systems to make notes that exist in the electronic medical record available to patients within password protected portals. A main impetus for Open Notes and My Healthe Vet was to promote a patient-centered approach to healthcare. Studies that have utilized these platforms have demonstrated patient benefits from their use. For instance, access to notes helped patients using My Healthe Vet adhere to prescribed medications, better prepare for clinic visits, and remember their plan of care[7]. Other studies utilizing Open Notes, which promotes patients’ online access to their healthcare notes[8], had similarly positive findings, such as increased comprehension of information, lower uncertainty about their condition, and in the cancer setting, increased trust in the oncologist[9].

Physicians have expressed some concerns about patient access to their notes. Participant physicians in the foundational 2010 Open Notes study[10] stated that knowing patients could access and read their notes would cause them to alter their usual clinical practice when writing notes. For example, some providers worried that using common medical phrases like, “the patient denies” or “the patient appears SOB [short of breath]” would anger patients. Subsequent studies found that about 20% of physicians reported changing the way they wrote about potentially offensive terms[11], confirming initial concerns. These concerns are especially pronounced amongst oncologists who are treating patients with a life threatening illness because the absence of relevant terminology could undermine efficient communication between clinicians. On the other hand, some concerns expressed by physicians in the original Open Notes study[10], such as the potential for patient notes to lead to longer clinic visit and legal ramifications from incorrect diagnoses, have not been realized[12].

In the oncology setting, cancer poses significant health threats that demand effective health communication[13]. In patient-provider communication, prognoses, complications of therapy, life support preferences, and hospice are common topics associated with cancer[14]. Communication around these topics can intensify patients’ feelings of fear, stigma and uncertainty[15, 16]. It is possible that these characteristics, combined with complex medical terminology[17], can be reflected in notes, making it important to understand what is included in oncological notes and how oncologists go about writing them. Notes related to terminal illnesses may be different from notes in the primary care setting, in which Open Notes has mainly been studied. Although both types of physicians manage patients with terminal illnesses, it is likely that more cancer patients than primary care patients have the potential to view distressing notes comprised of very complex information related to life threatening conditions. Aside from reported behavior change, to our knowledge there is no study that tests whether the content of notes and writing style were modified based on the potential for patients to view notes.

Previously, we established baseline characteristics of oncologists’ notes[18] using the linguistic tool, Linguistic Inquiry and Word Count 2015 (LIWC)[19]. We found that prior to the adoption of Open Notes, most notes were written in a formal manner with a negative emotional tone, and few oncologists included stigmatized terms like “obese”[18]. We extended that analysis for the current study and aimed to determine differences in the frequency of linguistic categories as determined by LIWC, and writing style as perceived by oncologists through interviews after implementation of Open Notes. We hypothesized that oncologists’ post- implementation notes would differ from the baseline due to modifications in documentation, such as length and tone of the note, since they were now aware that patients had the capability to view notes. We also aimed to discover oncologists’ attitudes and perceptions to understand their approach to writing notes and opinions about the potential for patients to view notes. To achieve this aim, qualitative interviews were conducted at the same time that quantitative data were collected and analyzed.

2. Methods

2.1. Setting

This study took place at a National Cancer Institute-designated cancer center in central Virginia. Recruitment for this study occurred between May-September 2016. This study was approved by the local Institutional Review Board of Virginia Commonwealth University.

2.2. Participants

We recruited oncologists from three departments (hematology, radiation, surgery) by presenting an overview of the study at each of three department meetings. In total, 22 out of 40 oncologists who were in attendance at those three meetings expressed verbal willingness to participate at the time and provided their names and contact information for future contact about participation. We contacted all interested oncologists and after reviewing the study information, written consent was obtained from willing participants to participate in both the interviews and analysis of their notes.

2.3. Procedure

This study employed a concurrent triangulation design[20] with mixed methods involving qualitative in-depth respondent interviews[21, 22] and quantitative analysis to assess the change in linguistic characteristics of oncologists’ notes before and after Open Notes implementation.

Since outpatient provider notes became viewable to patients in June 2015, the baseline data was established using notes aggregated from January 2014 – May 2014. Notes from this time period, over a year in advance of Open Notes implementation, were extracted because we wanted to ensure that notes were free of any bias or knowledge about the program, since announcements about Open Notes were made in the months prior. Notes for the post- implementation data were aggregated from January 2016 – May 2016, about seven months after implementation.

We accounted for variations in the number of patients seen by oncologists, and also in the number of notes generated, by randomly selecting notes from each oncologist. Notes included in the study were selected based on the total number of notes available per oncologist to ensure a representative sample, as some oncologists had hundreds of notes available and others had as few as 22. We selected 100% of notes from oncologists with 25 total notes or less, and smaller percentages from oncologists with more than 26 to ensure that there was an equal representation of notes. The average number of notes selected per oncologist was 17 pre-implementation and 25 post-implementation. The post-implementation time period may have had a greater number of notes available due to varying oncologist caseloads and complexities of patients’ illnesses at the time of consultation. All de-identified notes were maintained in a spreadsheet that included an ID number associated with the oncologist, date of the note, oncologist’s department and the note’s section header. Typically, oncologists included main section headers in their notes: “Assessment”; ”Plan”; ”Interval history”; ”Impression”; “Free Summary”; and “Laboratory Tests.” Descriptions of laboratory tests were omitted from our analysis since they mostly contained acronyms and numerical ranges. The other headers were included because they often contained subjective interpretation and information written by the clinician: “Assessment” was the most frequently used category by oncologists, and therefore was included in most samples here.

In conjunction with the linguistic analysis of notes, a semi-structured interview guide was developed to gauge oncologists’ attitudes and perceptions of writing notes knowing that they were viewable to patients. To construct the guide, a member of the research team (JA) wrote a set of initial questions, which was reviewed and scrutinized by the rest of the research team to ensure it aligned with the goals of the study. After a series of modifications, consensus was reached on the structure and content of the questionnaire. The semi-structured interview was written with flexibility, so additional follow-up questions could be asked based on individual responses[23]. Questions inquired about general thoughts on the adoption of Open Notes, the intended audience of notes, and whether their approach to writing has altered. Table 1 contains sample questions. Interviews were conducted by a member of the research team (JA) and audio recorded. The Atlas.ti v. 8.1.30.0 [24] software was used to manage the verbatim transcripts.

Table 1.

Interview guide sample questions

In your opinion, what has been the impact on your patients from viewing their information found on the portal? If you have observed negative outcomes, can you give specific examples? If you have observed positive outcomes, can you give specific examples? Has either caused emotional reactions (e.g. expressing gratitude or dismay upon viewing information on the portal?

In what ways, if at all, has your process of informing patients of diagnosis or discussing treatment regimens changed now that the OpenNotes policy is in place?

How do you feel about the possibility of patients discovering errors in your notes? How has this affected the way in which you write notes and the speed at which you contact patients once results are known?

2.4. Data Analysis

2.4.1. Notes

Extracted notes were input into the computer program, Linguistic Inquiry and Word Count 2015 (LIWC)[25] to identify language patterns. LIWC is empirically verified and widely used in the social sciences, particularly to analyze medical language[26–31]. It calculates words that reflect different emotions, thinking styles, social concerns, and parts of speech within a given text. LIWC reveals how individuals think as they write about events, since language is a marker of cognitive processes, style and social integration[32]. The research team reviewed all of the 82 available variables within LIWC and discussed which would be most relevant based on the aims and scope of this study. For instance, variables, such as “commas”, “parentheses”, and “adverbs” were omitted. Selecting the most valuable variables for the study’s purpose is common among health-oriented studies utilizing LIWC [29, 33]. We included components of LIWC that we believed may affect the patient–provider relationship. For example, analytical thinking, which is important because patients prefer providers who communicate in clear, simple terms without technical jargon[34]. High scores indicate very formal writing, whereas low analytical thinking scores tend to indicate narrative language focused on personal experiences[35]. Similarly, emotional tone indicates clinicians’ expressions of hope and optimism, which have reduced patients’ anxiety and helped foster stronger relationships[36]. The higher the number, the more positive the tone. Numbers below 50 suggest a negative emotional tone. Other variables included clout and authenticity. Clout suggests that the author of the note has high expertise and confidence, whereas low scores suggest tentativeness. High authentic scores indicate a more honest, personal, disclosing text, but low scores suggest a guarded and distant discourse[37].

Other language dimension variables from LIWC were also included in analysis:

1. Word count: The raw number of words contained in a note to understand if oncologists shortened or extended the length of notes.

2. Words per sentence: The mean number of words within each sentence of the note.

3. Words greater than six letters in each unit of analysis (i.e., provider note). A previous study has demonstrated that physician letters written at inappropriate reading levels may make them inaccessible to patients with low health literacy[38].

4. Personal pronouns: The total number of personal pronouns within a note, such as “I”, “we”, “you”, “he/she” and “they”. Previous research has found that providers predominantly use partnership building language like “we” to reference the patient- provider relationship[39].

5. Anxiety: Words associated with anxiety, such as “worried”, “fearful” and “nervous”. Clinicians may document prognostic information such as life expectancy[40], which may include words associated with anxiety.

6. Affect: Under the psychological processes category, encompassing words like “happy” and “cried”. The type of language that is used when communicating with patients, particularly for different diagnostic groups, can have an impact on diagnostic procedures or language-driven interventions[41].

7. Positive emotion: Also under the psychological process category, positive emotion consists of words such as “love”, “sweet”, and “nice.”

8. Negative emotion: Examples of words making up this category are “hurt”, “ugly”, and “nasty.” Patients prefer physicians when fewer negative emotion words were used during health care visits[42].

9. Sadness: A sub-set of the negative emotion category, words include “grief” and “sad”.

10. Power: “Superior” is an example of a power word. Oncologists have high levels of authority and power, and notes may reflect this imbalance with the patient.

Descriptive analyses and mixed effects modeling were performed for the above variables of interest. Random coefficient models were used to investigate changes over time in the language dimension variables by accounting for both inter-oncologist variability and also for intra-oncologist dependence over time, achieved by including time as a fixed effect and oncologist and an oncologist-by-time interaction as random effects. This model allows both unique baseline language dimension means as well as unique trajectories over time for each oncologist. For this model, a significant time effect (or a significant time-by-oncologist interaction effect) would be indicative of temporal changes in the language dimension variables. Note that the random interaction term was removed if it was not statistically significant at the 5% level, meaning that a random intercept model was used with a fixed time effect and a random oncologist effect. For this reduced model a significant time effect would be indicative of temporal changes in the language dimension variables. All analyses were conducted using the GLIMMIX procedure in the SAS statistical software (version 9.4, Cary, NC, USA).

2.4.2. Interviews

Transcripts of the interviews were reviewed by the research team and independent, open coding occurred using the constant comparative method[43]. Consistent with this method, each member of the research team independently developed themes to represent the underlying meaning of the text. The research team conducted regular meetings during which identified themes were presented and discussed and consensus achieved. Initially, as the methodology requires, only a few transcripts were analyzed. Once an exhaustive analysis of this original data set was complete, further small samples of transcripts were analyzed[44]. The themes which emerged from these data were compared with those from the original data set, and if necessary, new thematic categories were defined. This process continued until no new themes emerged and saturation was achieved[45]. Axial coding[46], the systematic identification of patterns and groupings of codes, was then used to conduct the last round of coding, followed by discussions with the entire research team to discuss the significance of codes, compared with the LIWC quantitative preliminary results, since the data was analyzed concurrently.

3.0. Results

3.1. Demographics

Thirteen oncologists were enrolled in the study. Eight from hematology/oncology, four from radiation oncology, and one surgical oncologist. The average age was 47, slightly more participants were female (54%), and the majority were white (77%). The enrollment rate was 59%.

3.2. LIWC Analysis

A total of 535 notes from all 13 oncologists were analyzed (207 pre-implementation and 328 notes post-implementation). In the pre-implementation data, most notes were from medical oncologists (45%, n=93) and radiation oncologists (51%, n=106). Data for post-implementation had similar proportions, with 45% (n=146) of notes from medical oncologists, 53% (n=173) from radiation oncologists, and 3% (n=9) from surgical oncologists. A descriptive table of note header type pre- and post-implementation is summarized in Table 2, with ‘Assessment’ being the most frequently available note header type (44–45%) and ‘Free Summary’ being the least frequent (4–10%). Table 3 includes representative examples of notes pre- and post- implementation for each of the LIWC summary variables. Mixed effect modeling results are presented in Table 4. However, the random oncologist effect was significant in all models, implying that the baseline language dimension values varied between oncologists. Note that the time-by-oncologist random effect was statistically significant for each of the language dimension variables, indicating that the change (or lack thereof) in the language variable dimensions was fairly consistent across oncologists; hence the interaction term was removed from each model. There were no significant differences between pre- and post-implementation. Dimensions of analytic thinking, clout, words per sentence, and words greater than 6 letters remained high pre and post implementation, while emotional tone and use of pronouns (personal and impersonal) remained low.

Table 2.

Frequency and percentage of note type

	Assessment	Interval History	Plan	Impression	Free Summary
Pre-Implementation	92 (44%)	39 (19%)	35 (17%)	21 (10%)	20 (10%)
Post-Implementation	149 (45%)	66 (20%)	53 (16%)	47 (14%)	13 (4%)

Table 3.

Excerpts of summary variables

Summary Variable	Pre-Implementation Note	Score	Post-Implementation Note	Score
Analytic
	1. Discussion today with patient and family regarding her scans. The lung lesions are essentially stable. There is minimal to no changes of the disease. The liver lesions are much better. They are easy to measure and there has been an excellent response to treatment there. Given that information will plan on doing another 3 cycles of Taxotere and gemcitabine. May have some additional issues with how well her counts tolerate this. However, it is showing to work and that is good for her. Other options if this does not work would be temozolomide. She has had Adriamycin in the past, so we would need to check cardiac function if we wanted to try drugs such as Doxil, but there may be limited room to use this given the past history. 2. Pain seems to be under much better control which is good. That should give her more energy. 3. Hypertension is under good control which also makes it easier to both treat her and to provide supportive care at this time. 4. She is doing well. We will see her again in three weeks. Will arrange for her chemotherapy and then see her again with her next cycle.	29.3	Since she was last seen she suffered a stroke but fortunately has had no residual neurologic deficits. During the hospital stay she suffered median nerve damage which she reports occurred during placement of an IV line. She reports this is improving. She has been seeing a hand specialist and undergoing occupational therapy.	53.6
Clout
	We received a notification from the lab about the critical calcium value later in the day on 5/13/14. We had Mr.[Name] return to the clinic to be admitted to the medicine service for treatment and evaluation to determine the cause of this problem. At this point, multiple myeloma seems to be the leading candidate, given his hypercalcemia, renal insufficiency, anemia, proteinuria and bone lesions on plane films. Will follow.	85.2	[Patient Name] has recovered well from radiation treatment. Her bone pain has completely resolved. She is not experiencing any toxicities from radiation treatment that require intervention. The fullness in her left ear is intermittent. It may improve since she is on a treatment break from the Sutent. I recommended she discuss this at the time of her followup with Dr. [Name]’s office on [Date].	77.6
Authentic
	I reviewed Mr. [Name] CT of the chest with the patient and I showed him the large right upper lung opacity as a result of the radiation treatment. I also showed him the disk changes at T8 level that are most likely responsible for the current pain. I also discussed with him the pulmonary nodules that are at present indeterminate.	55.6	I discussed that it is unclear if the right parotid lesion is a primary parotid squamous cell carcinoma or metastatic disease from squamous cell carcinoma of the face. I explained that because of the large size of the mass, adjuvant radiation therapy will likely be indicated. Final recommendations will be made after review of the surgical pathology, but likely a 6 week course of radiation will be recommended. I reviewed the process of radiation treatment. I explained the side effects of radiation treatment to the head and neck. I explained that radiation treatment will be directed to the surgical bed. He was accompanied by his wife and daughter. They had questions and their questions were answered to their satisfaction. He will return after surgery for review of the pathology and final radiation recommendations.	16.0
Emotional Tone
	Ms. [Name] returns to the clinic for routine followup visit. She was last seen in [Date]. Ms. [Name] reports that she has fair appetite and poor energy. She has the impression that she has lost some weight. However, she has no scale and has not been able to measure that. However, she was buying new clothes as her old ones were becoming too large. She denies any hemoptysis or shortness of breath, but continues to have dyspnea, unchanged from previous visit. She also has some cough with yellow sputum, but no fevers. She denies any dysphagia or esophagitis. Ms. [Name] comes in with a cast on her left leg. She explains that about 3 weeks ago she noticed a black spot on her left big toe that turned out to be an ulceration. Dr. [Name] is currently taking care of this lesion. Ms. [Name] reports that she is often constipated and is taking Senokot on a regular basis, otherwise she has bowel movements only once per week. She continues to have pain mostly in her left back and flank as well as her lower back, and knees with a history of arthritis. She was recently seen by Dr. [Name] from Pulmonology and continues on several inhalers.	34.1	On today’s visit, Ms. [Name] is represented by Dr. [Name], from Medical Oncology to discuss treatment for the single brain lesion in the right frontal brain. Ms. [Name] reports that overall she has been doing quite well. She denies any headaches or neurologic symptoms. She has good strength. Denies any seizures. Starting with a chemotherapy she felt more tired. She can manage nausea and her breathing overall has improved. Her major problem right now is pain in the lower L-spine for which she is taking oxycodone 5 to 15 mg as well as Aleve. She otherwise has regular bowel movements and no other specific problems.	42.9

Table 4.

Mixed effects modeling accounting for variation in oncologist and over time

					Random Effect
	Mean Pre-Implementation (SD)	Mean Post-Implementation (SD)	T (df = 22)	P-value	χ2 (df = 1)	P-value
Word count	133.6 (103.27)	104.5 (81.85)	−1.47	0.16	31.3	<0.0001
Analytic thinking	77.5 (21.50)	84.1 (17.85)	1.19	0.25	66.6	<0.0001
Clout	64.8 (18.12)	62.5 (17.98)	−0.61	0.55	36.9	<0.0001
Authenticity	25.1 (28.66)	30.5 (28.55)	1.0	0.31	27.9	<0.0001
Emotional tone	36.1 (26.43)	32.9 (25.97)	−0.7	0.47	10.9	0.0005
Words per sentence	15.1 (6.42)	16.0 (9.70)	0.9	0.38	96.0	<0.0001
Words >6 letters	26.9 (7.91)	28.8 (8.57)	0.43	0.67	34.7	<0.0001
Pronouns	8.8 (5.28)	6.6 (5.54)	−1.66	0.11	32.3	<0.0001
Personal pronouns	6.2 (4.14)	4.9 (4.62)	−1.23	0.23	31.8	<0.0001
‘We’ pronouns	0.9 (1.47)	0.5 (0.94)	−1.77	0.09	64.5	<0.0001
Affect	3.7 (4.31)	3.8 (3.99)	−0.30	0.77	36.3	<0.0001
Positive emotion	1.8 (1.72)	1.8 (1.96)	0.32	0.76	11.9	0.0003
Negative emotion	1.8 (3.90)	2.0 (3.00)	−0.57	0.57	57.5	<0.0001
Anxiety	0.29 (0.67)	0.32 (1.11)	0.57	0.59	4.0	0.02
Sadness	0.33 (0.70)	0.43 (1.26)	0.86	0.40	9.7	0.009
Power	1.76 (1.76)	2.15 (2.20)	1.58	0.13	7.85	0.0025

3.3. Interview Themes

All semi-structured interviews were conducted between April and September of 2016 in private offices, with the exception of one phone interview. The interviews averaged a half-hour in length. The three main themes that emerged were 1) Oncologists are cognizant of patients’ access to notes, 2) the intended audience for notes is not patients, and 3) the potential to generate interaction and communication.

3.3.1. Oncologists are cognizant of patients’ access to notes

Oncologists were keenly aware that patients logging into patient portals could easily view their notes, which was viewed as problematic. Although post-implementation LIWC results indicated that notes were largely unaffected, seven of the 13 oncologists interviewed reported that they were likely to slightly modify their style of note writing. A radiation oncologist stated that she had not consciously changed her note writing, but acknowledged that she ponders about the subject of patient access when writing the note. Similarly, a medical oncologist said that she “paid more attention to the notes” to ensure their accuracy, but stopped short of changing them because it would require writing “so many more words.” This paradox of trying to make the note less intimidating to patients while also limiting the word count was echoed by several oncologists. For example, an oncologist preferred to use less medical terminology so the note was “more clear and a little more patient friendly in terms of the language”, but found it difficult since notes require specific terms for billing purposes. Another medical oncologist recognized that he has become “guarded” because notes “can be misinterpreted as being judgmental or harsh when really you’re just trying to be effective and succinct.” He continued, “It severely limits your freedom to [frankly] describe what’s going on in your clinical note because normally you knew patients could get notes, but they had to go through a process.” The ease and availability of notes has resulted in “being a little more careful in terms of the wording,” according to another oncologist.

3.3.2. The intended audience for notes is not patients

In spite of oncologists’ concerns towards patients, there was widespread agreement that notes should be primarily written for colleagues, “To [ensure] that if another provider had to take care of that patient, they know what’s going on.” When asked who notes were written for, all thirteen oncologists replied by saying either “colleagues” or “myself and colleagues.” A medical oncologist immediately replied, “My nurse practitioners” because “We need to know what the game plan is, so that the next time we the patient, we’ll know where we’re going.” A radiation oncologist flatly stated, “The notes are not primarily written for the patient.” The use of notes for colleagues was exemplified by a medical oncologist who provided the following example: “You want to make sure you communicate to other providers if the patient shows up in the emergency room…if they end up going through a surgery, the surgeon needs to know their medications.” Prioritizing the content of notes with other providers in mind remained oncologists’ chief focus. A medical oncologist declared, “If a patient is non-compliant, I’ll put that in the note.” It was necessary for notes to remain unchanged because oncologists worried that edits could compromise the quality of the note. A medical oncologist said, “If your notes become wordier, they’re less likely to be read fully.”

3.3.3. Potential to generate interaction and communication

Less than half of oncologists interviewed (n=6) reported instances when patients raised concerns about the content of their notes. Examples of patient objections included corrections about a patient’s birthplace, labels such as “anxious” or “fatigued”, and use of the phrase “advanced directives.” The mentioning of “advanced directives” alarmed the patient because he was not sure what it meant. In retrospect, this particular oncologist would write the note the same way, but he and other oncologists were open to the opportunity of discussions with patients based on the content of notes. The possibility of patients raising questions about what is contained in the notes “could potentially lead to a better understanding of what’s going on with the patient.” A radiation oncologist recalled, “I’ve met very few patients that have been able to look at a pathology report and understand it from a surgery.” He continued, “They’ll almost always have questions if they meet with me…and that’s an in-person discussion.” In this manner, what originates as a patient concern or question could spark a conversation that might lead to improved care and better patient-provider communication. Although oncologists were reluctant to fully embrace Open Notes, most saw the potential in how patients could benefit. A medical oncologist acknowledged that patients viewing notes before visits allows them “time to process,” which can contribute to more productive consultation. Another radiation oncologist said, “I think it does add a little bit to the patients understanding because they can go back and see the options we talked about…[patients] being able to refresh their memory in terms of discussions gives them some advantage.” Lastly, an oncologist optimistically stated that if patients were more educated about Open Notes and understood the “intention of notes,” the potential for patient anxiety could be prevented.

4.0. Discussion and conclusion

4.1. Discussion

As recently as 2013, the Association of Medical Directors of Information Systems (AMDIS) proposed principles for electronic documentation, such as accuracy, meeting regulatory guidelines, and improving clinical decision support [47]. However, these principles do not explicitly address the possibility and ramifications of patients accessing such notes. Patients accessing provider notes can be considered as a form of pseudo communication, and the way in which notes are written may influence how patients perceive of their health care. Our study analyzed quantitative data from oncologists’ notes before and after Open Notes implementation, while also conducting a qualitative analysis to understand attitudes and behaviors of note writing.

Qualitative interviews revealed that oncologists in our sample were reluctant to change the content of notes. Patient portals allowing access to the electronic health record were designed to engage patients in their care and enhance patient centeredness[48, 49]. A study examining VA patient portal users found that additional clinical training was needed to emphasize patient- centered documentation practices[7]. Our findings uncovered that despite oncologists’ hesitancies towards Open Notes, they did recognize that notes can be more “patient-friendly” since patients will have access. Incorporating empathic language in medical encounters is widely accepted as an essential aspect of effective communication[50, 51] and has been associated with increased patient satisfaction[52]. Perhaps as more oncologists realize that their notes are being read by patients, they will modify their language to mimic the best practices they are already incorporating in face-to-face communication with patients.

LIWC analysis reinforced the qualitative findings and discovered no significant differences between oncologists’ notes written before and after implementation of Open Notes. It was hypothesized that oncologists would modify their writing behavior knowing patients would be monitoring it. However, similar to other studies that observed clinician behavior from audio or video recording during consultations, clinician performance and writing behavior did not change[53, 54]. One of the main reasons oncologists were resistant to altering their writing style was due to their goal of prioritizing accurate documentation to assist other providers caring for the patient. Multidisciplinary care, the collaborative treatment of patients by various specialists, has emerged as the standard of care in cancer management[55, 56]. The introduction of electronic health records has improved multidisciplinary team communication by allowing all practitioners involved in the treatment of a patient to document and view chemotherapy schedules, the goals of therapy, and treatment notes[57]. Although our findings indicate that oncologists placed a high level of importance on accurately documenting clinical notes in the EMR, electronic documentation has previously been criticized for overuse of copy and paste functions, which can perpetuate inaccurate medical information[58]. Moreover, repetitive notes are of little interest to the reader, patients or providers, and may obstruct important new data[59].

The language used in documentation for colleagues also has the potential to shape providers’ attitudes towards patients. A study that incorporated both neutral and stigmatized language into primary care notes, such as casting doubt on the patient’s pain and negative portrayals of the patient, found that medical students and residents formed negative attitudes towards the patient when exposed to the stigmatizing language[60]. The notion that the linguistic elements used in notes matter is widely documented among other Open Notes studies. Similarly, sharing notes has resulted in perceived relational benefits to both providers and patients, and may be particularly helpful in elevating patients’ perceptions of their provider within vulnerable patient populations[61]. The mechanism for reading provider’s notes is now being expanded to enable patients to directly input feedback into their note. Patients participating in pilot projects have the ability to prioritize concerns by sharing their agenda with providers[62], and have reported greater engagement in their care and the desire to help clinicians improve note accuracy[63].

Limitations of the current study include it being confined to one health system and therefore, may not be generalizable outside of an academic cancer center or to other academic health institutions with unique system implementation processes. Moreover, the writing style of radiation and surgical oncologists may differ from medical oncologists, and thus may not be fully represented in the low sample size of these specialties. It may also be possible that other factors could have impacted the results, such as history threats to validity, as all changes can’t be attributed to the implementation of Open Notes. Any inference regarding the results should be tempered in the light of these potential factors, although we are not aware of any. Similarly, we analyzed notes from a year prior to the Open Notes implementation, but it is possible that other system changes occurred during that period which could have affected oncologists’ notes. It is possible that randomly selected post-implementation notes contained a greater number of more complex cases than pre-implementation data. Additionally, participating oncologists were self- selected and their views may differ from oncologists who chose not to participate or did not attend the department meeting. Finally, patient portal adoption and electronic access to notes is still in its nascent stage. Therefore, oncologists may not have adjusted their writing style because most of their patients may not be enrolled in the portal.

The majority of Open Notes studies have focused on the primary care setting. Our study highlighted the cancer care delivery process, which may present unique challenges that may not occur in other settings, since terminal illnesses are common and topics like prognoses and hospice are often discussed[14]. The results from the current study, as well as the baseline study, indicate that oncologists worry about patient’s access to notes due to the sensitivity and complexity of cancer[18]. Future studies should seek further understanding regarding the impact open access to provider notes has on the content of notes and patient-provider communication, particularly in regards to disclosing diagnoses.

As patients become more accustomed to having access, longitudinal studies can examine whether reading notes motivates patients to be more empowered about their health and engaged in their care, such as asking questions, participating in shared decision-making and using notes to initiate discussion with other providers. Additionally, the writing behavior of providers should continue to be monitored to determine whether the quality of notes improves or declines.

4.2. Conclusion

Our study revealed that oncologists primarily did not intend to change their notes because they believed notes were intended for other providers. A linguistic analysis between pre- and post-implementation of notes confirmed these views and found that notes did not appreciably change.

4.3. Practical implications

As more patients adopt portals and gain access to notes, clinicians should be aware that patients may have the capability to view them. Easy access to notes can serve as an opportunity for clinicians to reinforce and clarify important aspects of the consultation in language that patients can understand.

HIGHLIGHTS.

• Pre-and post- implementation of oncologists’ clinic notes were analyzed

• No significant differences were found in note content pre- and post- analysis

• Oncologists were hesitant to embrace Open Notes, but acknowledge potential benefits