Clinical Notes Contain Limited Documentation of Shared Decision Making for Colorectal Cancer Screening Decisions

Abstract

Background.

Effective shared decision making (SDM) in health care involves thorough discussions of options, pros, cons, and patient preferences. While SDM is recommended for engaging adults aged 76 to 85 y in colorectal cancer (CRC) screening decisions, the extent of SDM documentation in clinical notes remains unclear.

Objective.

This study aimed to evaluate the current state of SDM documentation in clinical notes regarding CRC screening discussions for adults aged 76 to 85 y. It also sought to assess the impact of an SDM training intervention on documentation quality and compare documented SDM elements with physician- and patient-reported SDM.

Methods.

Data from 465 patient participants and 58 primary care physicians in a multisite cluster randomized trial were analyzed. Physicians in the intervention arm underwent a 2-h SDM skills training and received support tools, including an electronic health record SmartPhrase. Coders analyzed clinical notes using content analysis to identify SDM elements. Linear multilevel models and multilevel partial correlations were used for analysis.

Results.

Overall, SDM Note scores were low ($\overline{x}$ = 0.80, s = 0.99). The intervention arm exhibited higher SDM Note scores than the comparator arm did (adjusted mean 1.02 v. 0.66; P = 0.006), with more frequent documentation of stool-based tests (52% v. 33%; P = 0.02) and colonoscopy cons (28% v. 8%; P = 0.001). No significant differences were observed in patient preference documentation. SDM Note scores correlated moderately with patient- and physician-reported SDM.

Conclusion.

Documentation of CRC screening discussions with older adults lacks comprehensive SDM elements. The intervention improved SDM documentation, particularly regarding alternative screening options and potential cons. Given the limited documentation of SDM even after a training intervention, attention to more robust SDM documentation, including patient preferences and discussion of stopping CRC screening, is needed.

Introduction

Shared decision making (SDM) has increasingly found its way into the mainstream as health systems continue shifts toward patient-centered care models.^1–5 Patients are encouraged to engage in individual advocacy when making treatment decisions, and insurers are more often requiring SDM as a condition of coverage for a variety of procedures and tests.^2,4–8 One example is the Centers for Medicare & Medicaid Services (CMS) requirement for patients to have completed an SDM conversation prior to the use of low-dose computed tomography for lung cancer screening (LCS).^2,9 The use of clinical notes as a record of SDM conversations has become pertinent for reimbursement as a result.^10–12

CMS specifies the necessary components of SDM conversations for reimbursement purposes—such as the conversation supported by the use of decision aids, the discussion of benefits and harms of screening, follow-up tests, overdiagnosis, and false positives; however, CMS language remains vague regarding whether and/or how these elements should be documented.¹¹ For instance, the CMS Web site specifies that to secure coverage for left atrial appendage closure, physicians must “document shared decision-making discussions correctly.”¹³ However, the CMS Web site leaves uncertainty about what constitutes adequate documentation for these decision-making conversations contributing to variability in documentation and discordance with CMS and other regulatory oversight guidelines.¹¹

The lack of standardization makes it difficult to understand the quality of SDM conversations and, more importantly, how this quality is related to perceived SDM on the part of the patient.^6,14 The significance of high-quality SDM documentation in clinical notes extends beyond reimbursement considerations. The adoption of open-note programs has increased patient access to their clinical notes, emphasizing the need for clear and thoughtful documentation of patients’ decisions and values.^4,6,15,16 In addition, well-documented clinical reasoning and conversations with patients can offer malpractice protections for clinicians.^6,10,12,17 Furthermore, comprehensive documentation has the potential to ensure continuity of care by communicating the rationale behind medical decisions to other health care providers.^7,16

The literature on SDM documentation in clinical notes, although sparse, often describes SDM documentation as limited and highly variable in relation to specific SDM items—such as options, alternatives, risks, and benefits—documented in clinical notes.^8,11,15 For example, although CMS requires a documented SDM encounter before covering an LCS examination, only 42% of individuals undergoing LCS have SDM documented in their electronic health record (EHR).¹¹ Of the 42% of EHR records containing SDM, only one-fifth contained documentation of all required CMS components of SDM.¹¹ In a large cohort of patients eligible for an implantable cardioverter-defibrillator, discussions about the risks and benefits of implementation were documented in fewer than 50% of cases.¹⁸ In a study assessing the documentation of SDM in diagnostic testing for dementia, findings revealed that SDM was poorly documented, particularly regarding options discussed and final decisions made.¹⁹ Literature points to the lack of provider training, insufficient information systems, and, importantly, challenges in documentation as regular barriers to SDM implementation in clinical practice.^10,20 These findings underscore a potential gap between SDM conversations conducted across various medical specialties and the thorough documentation of those discussions.

We sought to understand the current state of SDM documentation in clinical notes and explore how clinical note documentation reflects SDM conversations as perceived by patients and clinicians. We chose to explore SDM documentation in colon cancer screening (CRC) decisions for older adults given that the US Preventive Service Task Force (USPSTF) recommends the use of SDM to support decisions about continuing CRC screening for adults aged 76 to 85y.^21,22 Therefore, our group assumed that some level of SDM documentation may exist in clinical notes on this topic. Although documentation of SDM is not currently required by CMS for CRC screening reimbursement in this older adult population, high-quality documentation of this decision is useful for ongoing clinical care given the nuances of the clinical circumstances. Thus, it is important to begin to explore the state of clinical documentation and experiences of SDM in this area, as well as to identify interventions that may improve these practices. This study examines the documentation of SDM for CRC screening decisions for older adults in clinical notes, evaluates whether physicians who received SDM training have more SDM documentation in the clinical note than physicians who did not receive SDM training, and compares SDM documentation to patient- and physician-reported SDM.

Methods

This study is a secondary analysis of data from the PRIMED study, a multisite cluster randomized trial (conducted October 2019 to April 2021) evaluating SDM in CRC screening decisions for older adults aged 76 to 85 y.²³ In PRIMED, physicians were randomized to intervention or comparator arms. Physicians in the intervention arm of the trial completed a 2-h online course in SDM communication skills including video vignettes with examples of SDM communication in the context of CRC screening decisions for adults 76 to 85 y old. In addition, physicians were provided patient education materials and a SmartPhrase to document SDM discussions in the clinical note. The SmartPhrase included prepopulated text with the latest USPSTF recommendations for CRC screening among patients 75 to 85 y, text describing 3 screening options (colonoscopy, stool-based test, no further screening), and space for clinicians to indicate the amount of time dedicated to the discussion, other aspects of the discussion if applicable, and finally the patient’s preferred screening method. Physicians also received an electronic reminder via e-mail or EHR message per physician preference 2 to 3 d prior to the patient’s visit encouraging them to have a conversation with the patient about CRC screening for patients who were due or past due for screening. The comparator arm of the study received electronic reminders only without any SDM training. Additional details on the SDM training and how the final sample was derived can be found in the published manuscript for the original PRIMED study.²³

Physicians were enrolled from primary care practices at 3 academic medical centers and 2 community hospitals in New England. Patients were eligible for the study if they were 76 to 85 y old, due for a discussion about CRC screening, attended a nonurgent medical visit within the study period with a participating physician, able to read in English or Spanish, and could provide consent for themselves. Patients were deemed ineligible if they had a history of colon or rectal cancer, prior colectomy, or were identified by their physician as having an exclusionary comorbidity (e.g., active malignancy, end-stage organ failure) or limited cognitive ability. Study staff screened for eligible patients among the enrolled physicians prior to their scheduled visit to reconfirm eligibility and to invite patients to participate.

Patients and physicians were asked to complete a post-visit survey evaluating conversations about CRC screening using the SDM Process Scale (for patients) or an adapted version of the SDM Process Scale (for physicians). The SDM Process Scale is a 4-item measure with scores ranging from 0 to 4 (with higher scores indicating more SDM).²⁴ The scale measured whether clinical conversations included discussion of no further screening, pros and cons of screening, and whether or not they discussed the patient’s treatment preference. This scale has been validated as a patient-reported measure across a variety of clinical areas including preventive health and cancer screening.^20,21 The physician version of the scale was created from the validated patient version to ask items from the physician’s perspective (e.g., “How much did you talk about the reasons…” instead of “How much did you and your doctor talk about the reasons…”).

The clinical note associated with the visit was abstracted for each surveyed patient from the her, and variables of interest were identified to create a codebook (see Appendix A). The codebook was based on items associated with the SDM Process Scale and physician feedback.²⁴ Two study staff (J.S. and B.M.) double-coded all variables independently and reviewed coding for agreement after about every 90 notes. Kappa value and percentage agreement were assessed (threshold of kappa ≥0.70 or agreement >70%) to ensure coders interpreted and applied the agreed upon codes to the data similarly.²⁵ Documents were recoded until kappa or agreement values met the threshold established a priori, and then all disagreements were resolved following discussion among coders. An example of kappa values and percentage agreements reached prior to final resolution can be found in Appendix B.

All notes were evaluated for documentation of CRC screening discussion (yes/no). Among the notes in which a discussion of CRC screening was documented, study staff coded for the presence of 4 SDM items associated with the SDM Process Scale, including whether the physician documented any CRC screening options (options consisting of colonoscopy, stool-based tests, and no further screening), whether there was documentation of the pros and cons of each aforementioned option, and if the patient’s treatment preference was documented. Each of the items was coded individually as yes/no (e.g., colonoscopy documented [yes/no], colonoscopy pros [yes/no], colonoscopy cons [yes/no], stool test documented [yes/no], etc.). Excerpts from the clinical note containing this documentation were extracted for contextual support when coding. Each code and 2 example excerpts per code are provided in Table 1. Examples are representative of the variation in documentation in the clinical note and include first a generic example in which a single keyword was included in the document (e.g., “benefits” for the pros code) followed by a more detailed example that provided more specific information about the coded variable (e.g., “able to detect both premalignant polyps and malignancy” for pros).

Table 1.

Clinical Note Coding and Scoring Examples

Code	Excerpt from Clinical Note
Pros	“Discussed risks and benefits of doing colonoscopy.” “Colonoscopy: Pros: able to detect both premalignant polyps and malignancy, done less frequently (interval depends on findings).”
Cons	“We have discussed the pros and cons of continuing colon cancer screening at this time.” “She is also concerned about risks of colonoscopy … and potential inability to complete a bowel prep.”
Options	“Discussed various other options.” “Discussed there are two colon cancer non-invasive screening tests, Cologuard and FIT test to be performed at home.”
Preference	“Prefers to go for colonoscopy.” “He reaffirms his previous expressed belief that he does not want any further testing.”

FIT, fecal immunochemical test.

An SDM Note score was calculated by assigning points to each SDM item present in the clinical note (options, pros, cons, and preferences) and summing for total score ranging from 0 to 4. We chose to align the SDM Note score as closely with the SDM Process Scale items, which allots a total of 1 point for provision of options, a total of 1 point for provision of pros, a total of 1 point for provision of cons, and a total of 1 point for discussion of preference. Thus, SDM Note scores were composed of 1 point for discussing no screening as an option, one-third point for discussing a benefit of each of the 3 screening options (colonoscopy, stool-based test [SBT], no screening), one-third point for documenting a risk for each of the 3 screening options, and 1 point for documentation of patient preference. Higher scores indicate more complete documentation of SDM. Clinical notes that contained no SDM items or did not contain any mention of CRC screening were scored as 0.

Analysis

We report descriptive statistics related to patient and physician demographics and the SDM elements documented regarding colon cancer screening in the notes.

We analyzed the following 2 research questions:

1. Did physicians in the intervention arm have higher SDM Note scores than those in the comparator arm?

2. How does SDM documented in the clinical note relate to patient reports of SDM and physician reports of SDM?

For the first research question, linear multilevel models were used to evaluate the differences in SDM Note scores between intervention and comparator arms while adjusting for the nesting of patients within physicians. Logistic multilevel models were used to evaluate whether those in the intervention arm document specific SDM items more often than the comparator (pros, cons, options, preferences). These analyses used the full patient sample.

To address the second research question, multilevel partial correlations were used to evaluate the similarities between patient-reported SDM and SDM Note scores as well as physician-reported SDM and SDM Note scores. Sensitivity analyses were conducted to identify if the study arm affected these relationships and to identify if the lack of discussion of CRC (either in the note or in the visit as reported by patient or physician) affected these relationships. As no differences in conclusions were found between models, we report on those models not adjusting for study arm and using all data regardless of whether CRC was discussed in the note or the visit. These analyses used the subsample of patients in which all data were available: clinical note, patient-reported SDM, and physician-reported SDM.

Results

Of the 465 visits, 236 (50.8%) occurred with physicians in the intervention arm and 229 (49.2%) occurred with physicians in the comparator arm. Physicians (n = 58) were evenly distributed by gender, with an average of 22 (s = 10) y in practice, and generally limited prior SDM training (n = 16, 28%). Additional physician sample characteristics are found in Table 2. Patients were on average 79 y old (s = 3 y) and slightly more female (53%) than male. Most were White non-Hispanic (93%), highly educated (34% with more than a 4-y college degree), reported being in either good or excellent physical health (51%), and displayed high health literacy (83%). More than half of the patient sample reported having had a prior colonoscopy (67%).

Table 2.

Physician Characteristics (n = 58)

Characteristic
Age, y, $\overline{x}$ (s)	53 (9)
Female, n (%)	29 (50)
Years in practice, $\overline{x}$ (s)	22 (10)
Number of completed surveys, median (range)	7 (4–11)
Prior shared decision-making training, n (%)	16 (28)
Academic practice (v. community practice), n (%)	37 (64)

Most (70%, 326/465) notes included documentation of a CRC screening discussion. Table 3 describes the presence of SDM elements among all clinical notes that discussed CRC screening. Of those notes that discussed CRC screening, most (98%) discussed screening tests, with colonoscopy being the most common option listed, followed by stool-based tests and, finally, no screening. For all testing options, documentation of pros and cons was severely limited, although cons were documented more often than pros were regardless of which option was being discussed. Patient preference for CRC screening was documented in 62% of notes.

Table 3.

Presence of Shared Decision-Making Elements among 326 Clinical Notes where Colorectal Cancer Screening Discussions Were Documented

Code	Percentage of Notes with Code
CRC screening tests	98% (319)
Colonoscopy	85% (272)
Stool-based test	44% (143)
No screening	30% (98)
CRC screening pros
Colonoscopy pros	16% (53/326)
Stool-based test pros	9% (28/326)
No screening pros	5% (16/326)
CRC screening cons
Colonoscopy cons	19% (62/326)
Stool-based test cons	12% (40/326)
No screening cons	6% (20/326)
Patient preference for CRC screening	62% (202/326)

CRC, colorectal cancer.

Research Question 1: Did Physicians in the Intervention Arm Have Higher SDM Note Scores Than Those in the Comparator Arm Did?

The SDM Note scores (scale: 0–4) were low overall ($\overline{x}$ = 0.80, s = 0.99). The SDM Note scores in the intervention arm were higher (adjusted mean = 1.02, SE = 0.1) than those in the comparator arm (adjusted mean = 0.66, SE = 0.09; P = 0.006, intraclass correlation coefficient = 0.11). Some differences in documentation were found by study arm (Table 4). Although more notes in the intervention arm contained CRC screening documentation compared with the comparator arm (76% v. 66%), this difference was not statistically significant when adjusting for the nesting of patients within physicians. Significant differences were found between intervention and comparator arm documentation of SBTs and specifically SBT pros and cons (see Table 4). The intervention arm also documented the cons of colonoscopy more frequently than the comparator arm. Although the intervention arm documented the pros and cons of colonoscopy and no screening options more frequently than the comparator arm, these differences were not significant (P > 0.05).

Table 4.

Comparison of Adjusted Percentage of Shared Decision-Making Items among Comparator and Intervention Arms

	Comparator Adjusted % (95% CI) n = 229	Intervention Adjusted % (95% CI) n = 236	P Value	ICC
Colorectal cancer (CRC) discussion documented	66% (56%−75%)	76% (66%−84%)	0.14	0.18
One or more CRC option documented	93% (90%−98%)	100 % (0%−100%)	1.00	<0.01
Colonoscopy	82% (70%−90%)	87% (76%−93%)	0.49	0.32
Stool-based test (SBT)	33% (23%−44%)	52% (41%−63%)	0.02	0.17
No screening	25% (18%−34%)	36% (27%−45%)	0.09	0.07
CRC pros documented
Colonoscopy pros	11% (5%−20%)	20% (12%−32%)	0.13	0.30
SBT pros	5% (2%−10%)	12% (8%−19%)	0.03	0.08
No screening pros	2% (0.4%−6.3%)	6% (2.6%−14.6%)	0.10	0.44
CRC cons documented
Colonoscopy cons	8% (4%−16%)	28% (19%−39%)	0.001	0.20
SBT cons	7% (3%−13%)	16% (10%−26%)	0.04	0.23
No screening cons	2% (0.7%−7.8%)	8% (3.6%−16.6%)	0.09	0.38
CRC preference	60% (49%−70%)	66% (55%−72%)	0.41	0.14

CRC, colorectal cancer; SBT, stool-based test.

Research Question 2: How Does SDM Documented in the Clinical Note Relate to Patient Reports of SDM and Physician Reports of SDM?

Of the 465 patient visits, 382 contained all available data (clinical note, patient-reported SDM, and physician-reported SDM) and were used in analyses comparing SDM Note scores to patient- and physician-reported SDM. Overall, both patient-reported SDM Process Scale scores (scored 0–4; $\overline{x}$ = 1.27, s = 1.23) and physician-reported SDM Process Scale scores ($\overline{x}$ = 1.39, s = 1.24) were relatively low. SDM Note scores were significantly correlated with both patient-reported (r = 0.28, 95% confidence interval [CI] [0.19–0.36], P < 0.001) and physician-reported (r = 0.42, 95% CI [0.33–0.50], P < 0.001) scores.

Discussion

Despite low SDM scores, a key strength of this study was in obtaining 3 different measures of SDM (physician, patient, and clinical note) reflecting the visit. All 3 measurements indicated low SDM occurrence (i.e., scores ranging from 0.8 to 1.4 on a scale ranging from 0 to 4). Even still, small to moderate positive correlations were found when comparing SDM Note scores to patient and physician scores (respectively), suggesting some relationship exists between the SDM documented in the note and SDM being perceived by the patient and physician during the visit. We see that SDM documentation, and the subsequent Note score, can be positively influenced by additional SDM training, as evidenced by the significantly higher SDM Note scores in the intervention arm than the comparator arm. We are limited in our understanding of the magnitude of the SDM Note score given the lack of standards in note documentation more broadly and the novel use of the SDM Note score in the study. However, others have reported on the usage of the SDM Process Scale in other clinical areas, including cancer screening.^24,26–28 Vo and colleagues²⁶ reported SDM Process Scale scores ranging from 1.24 to 1.94 when used to evaluate cancer screening decisions and 1.40 to 2.65 when evaluating medication decisions. In comparison, the SDM Process Scale scores for physicians and patients resulting from this study fall among the lower range of scores reported by the literature using the measure in other clinical contexts.^24,26–28

Our overall findings are similar to those produced by Tailor and colleagues,¹¹ who also found differences in SDM reported by patients compared with those documented in clinical notes, where patients perceived SDM occurring more frequently than had been documented in the note (71% versus 42%). With this discrepancy, it is difficult to ascertain whether physicians are underdocumenting SDM or patients are overestimating SDM in their reports. Researchers in SDM measurement science have noted a benefit of using dyadic measurement—assessing both patient and physician perspectives—to get a more complete picture of SDM.²⁹ Regardless of whether we consider the clinician-documented note or patient- or physician-reported SDM, considerable improvements in SDM conversation documentation are needed, as evidenced by our data.

Overall, physician documentation of SDM discussions regarding CRC screening in older adults was quite limited. While most physicians who documented a CRC screening discussion included documentation of colonoscopy as a screening option, fewer physicians mentioned stool-based tests or no further screening as options for patients. A national survey conducted by the Centers for Disease Control and Prevention (CDC) found that primary care doctors and gastroenterologists consistently advocated for colonoscopy as the primary CRC screening method for average-risk patients, followed by various stool-based tests.³⁰ This trend indicates that despite the availability of approved and recommended stool-based tests, colonoscopy remains the preferred test among US clinicians. The CDC’s survey also looked at factors associated with clinicians routinely recommending specific CRC screening options. For SBTs, familiarity with the method, perceived effectiveness of the method, and Medicare coverage knowledge all increased the odds of clinicians offering SBT as a screening option for patients.^30,31 Although our study did not directly assess these factors, they could offer valuable insights into why SBTs were documented less often than colonoscopies in clinical notes.

Thorough SDM conversations about CRC screening should include the option of stopping screening.^30,31 However, our study found limited documentation of no further screening, which could indicate a potential reluctance to bring up the topic.³² Studies have documented the hesitancy of physicians to initiate conversations about cessation of screening tests for older adults.^30,33 One study examining screening cessation in adults with limited life expectancy identified that older adults were more open to stopping cancer screening within a trusting clinician–patient relationship.³⁰ However, even among those unlikely to benefit from screening, only one-third changed their minds after clinician discussions.³⁰ Patients with recent screenings varied in their willingness to stop, and some reacted negatively to recommendations to cease screening.³² Moreover, clinicians may be less familiar or less comfortable with USPSTF guidelines regarding possible CRC screening cessation for older adults aged 76 to 85 y. A CDC national CRC survey indicated that 27.7% of primary care providers (PCPs) reported that they did not have an upper age limit for CRC screening colonoscopy.³⁴ Ultimately, given the relatively low mention of no further screening as an option, documentation of CRC discussions in the clinical note demonstrated limited SDM for these conversations.

PCPs in the intervention arm had significantly higher SDM Note scores than those in the comparator arm did. The higher SDM scores observed in the trained group align with existing literature and support the findings of the larger PRIMED study, whose results showed additional SDM training led to increased SDM Process scores, thus reinforcing that training enhances proficiency in conducting SDM.²³ A systematic review of 29 SDM training programs demonstrated that such training leads to moderate improvements in both patient- and observer-reported SDM skills.³⁵ Higher scores in the intervention arm of our study were due to the greater inclusion of 2 elements: SBTs as an option along with its pros and cons and a greater frequency of mentioning the cons of colonoscopies. The PRIMED SDM training intervention explicitly addressed challenges inherent in conversations about deimplementation and encouraged PCPs to consider 2 options with patients 1) stopping the screening completely or 2) reducing invasiveness, in this case, switching from colonoscopy to stool-based tests.²³ Our results suggest that intervention arm physicians were more amenable to offering SBTs. Importantly, the ability of physicians to switch patients from colonoscopy to stool-based tests may be more in line with patient preferences.³¹ Recent findings indicate that 37.3% of adults older than 50 y express a preference for multitarget SBT over colonoscopy.³¹ In contrast, a mere 13.6% of adults in the same age group identify colonoscopy as their preferred CRC screening method.³¹ Clinicians’ ability to facilitate the transition from colonoscopy to SBT when clinically appropriate may enhance decision satisfaction, a crucial outcome in the realm of quality SDM.

PCPs in the intervention arm also exhibited a notably higher tendency to document the cons of colonoscopy compared with their counterparts in the comparator arm. Communicating the cons or potential harms of a treatment or screening test allows patients to make informed decisions aligned with their values and preferences, thereby fostering a more personalized and patient-centered approach to care. Ultimately, physicians who received formal SDM training did document more SDM elements than those in the comparator arm did, again suggesting that SDM training had a positive impact on documentation.

The SDM skills training intervention was guided by a conceptual framework designed to enhance clinical decision making, as outlined by Mulley and Sepucha et al.^36–38 This framework highlights 6 critical steps: 1) identifying when a decision needs to be made, 2) clearly presenting available options, 3) discussing the benefits and risks of each option, 4) eliciting and understanding patients’ preferences and goals, 5) facilitating thoughtful deliberation to determine the best course of action, and 6) supporting the implementation of the chosen decision. The training equips clinicians with practical strategies to apply these principles in their practice, aiming to improve the quality and consistency of SDM in patient care. This framework served as the foundation for the study’s primary intervention, which included an online SDM training course and EHR-integrated tools to prompt and support SDM discussions during clinical visits.³⁹

The logic model created by Sepucha and colleagues³⁹ for the PRIMED study activities outlines the relationships associated with the intervention and their impact on the decision-making outcomes. We believe clinical documentation affects the visit and the decision-making outcomes of the model by providing tangible evidence of key elements of the steps, such as the discussion of options, incorporation of patient goals and concerns, and documenting the patient’s preference.³⁹ Clinical note documentation taken together with the physician-reported SDM (representing self-assessment of efforts to engage the patient) and the patient’s perceived SDM (representative of their involvement in the SDM process) offers a more comprehensive view of the SDM process. The intersection of these components may provide a more accurate depiction of SDM during a clinical visit. Strong correlations between patient and physician perceptions, as well as alignment with the SDM documentation, validate the thoroughness of the process and its accurate representation in the clinical record. By analyzing these relationships, the study sought to identify opportunities to enhance SDM practices, ensuring they are both effectively executed and accurately documented.

The importance of good clinical documentation is widely recognized, as it plays a critical role in ensuring quality patient care.⁸ Effective documentation supports patient safety and education, serves as a safeguard in medical–legal matters, facilitates reimbursement processes, and enables clear communication of care plans both with patients and within the health care team.^{4,8,21,22,29,40} However, the data presented in this article suggest the need for a more accurate and consistent method of documenting SDM in clinical notes. This is consistent with a growing body of literature citing limitations in understanding potential relationships in documentation and care quality more broadly due to content variation resulting from a lack of standardization and clear definitions of what constitutes adequate documentation.^6,14 While some measures have been designed to evaluate the quality of clinical note documentation, to our knowledge, none have explicitly included an evaluation of SDM elements in clinical notes despite SDM’s increasing relevance in clinical care. Our study presents one method for evaluating clinical notes for SDM documentation via the SDM Note scores and suggests the documentation of SDM in clinical notes may represent the nature of SDM conversations as perceived by the patient and clinician with some accuracy. Furthermore, our results show that additional SDM training for physicians increases the chances these conversations will be more comprehensive (i.e., include more SDM elements including further treatment options and the pros and cons of those options), as seen in the higher SDM Notes scores among physicians who engaged in the training intervention.

Limitations

The patient sample represented a relatively homogenous group, with respect to race and ethnicity, from a small geographical region. The study focused on primary care physicians, and it is not clear whether the findings will generalize to other specialties or other decisions within primary care. The additional evaluation and comparison of physician and patient perceptions of SDM occurring using the SDM Process Scale will contribute to establishing a standard to which we can compare the magnitude of these outcomes within different clinical contexts. The further exploration of SDM documentation across a wider range of clinical areas may provide us a broader understanding of the landscape of SDM more globally. Despite our effort to support SDM-driven documentation by providing study physicians with a SmartPhrase, this tool was rarely used in practice. Future work would benefit from understanding the facilitators and barriers to using such a tool, considering the potential of this tool to create clinical notes that contain specific and prepopulated information for SDM or other contexts. In addition, as there was no audio recording of the interactions, researchers could not compare clinical notes to an observer-reported gold standard, which limits our ability to assess the accuracy of the documentation. Future studies would benefit from using such an approach to strengthen our understanding between what is documented in the clinical note and whether SDM conversations are occurring. It is important to recognize that while our analysis focused on a simple definition of SDM, more intricate definitions may offer additional insights into the complexities of patient–provider decision-making dynamics.

Conclusion

Physicians had limited documentation of SDM in clinical notes regarding their discussions about CRC screening with older adult patients. Formal SDM training increased documentation about testing options, including SBTs and the cons of colonoscopy. However, despite improvements in the intervention arm, these overall low levels of SDM documentation, even by physicians who had specific training in SDM, underscore the need to design clinician training, and EHRs, that more readily capture and document the SDM process.

Supplementary Material

Supplementary material for this article is available online at https://doi.org/10.1177/0272989X251340704.

Highlights.

• Shared decision-making (SDM) documentation in clinical notes is limited for discussions on colon cancer screening among older adults.

• SDM training improves SDM documentation of screening options for colorectal cancer, specifically documentation of stool-based testing and the downsides of screening options.

• SDM documentation in clinical notes is related to patient and provider reports of SDM.

Data Availability

Analytics code and research methods can be shared upon reasonable request to the corresponding author. Data can be shared upon reasonable request with completion of a data use agreement given the protected health information in the qualitative visit note data.