PSAzing up patient care: a quality improvement project for informed cancer screening

Abstract

Introduction

Prostate cancer is a leading cause of cancer-related deaths among men in the USA. Prostate-specific antigen (PSA) testing for screening remains controversial, with shared decision-making (SDM) recommended by professional guidelines to discuss screening’s risks, benefits and uncertainties. Despite these recommendations, SDM is underused, with only 10% of men receiving comprehensive SDM. This quality improvement project aimed to improve SDM documentation and the implementation of PSA screening in an urban safety-net, resident-led primary care clinic.

Methods

We implemented a continuity clinic note template with a specific SDM ‘dot phrase’ to improve the documentation and execution of SDM conversations. Our primary aim was to increase SDM documentation for prostate cancer screening, with a secondary aim to improve follow-up on abnormal PSA values. The intervention included men aged 55–69 years. Preintervention, residents were educated on SDM and PSA screening. Postintervention, patient charts were reviewed for documentation rates and screening outcomes. Feedback was collected during dedicated sessions. Finally, comparative statistics were conducted between baseline preintervention and eligible postintervention cohorts.

Results

SDM documentation improved significantly from 7.1% preintervention to 37.2% postintervention (p<0.001). PSA screening rates increased from 31.5% to 37.8% (p=0.155), though not significantly. Notably, 49.3% of patients declined PSA testing post-SDM, and 68.5% of previously screened patients were up to date with PSA testing. Residents reported challenges with SDM implementation, including time constraints and patient acuity.

Conclusion

Templated notes and dot phrases significantly improved SDM documentation, both compared with our clinic baseline rates and compared with recent reported national rates, overall enhancing standardised preventive care in primary care. Although PSA screening rates improved, challenges such as time limitations and patient no-shows impacted the intervention’s effectiveness. Future cycles will address these barriers to improve outcomes further.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Prostate cancer screening using prostate-specific antigen (PSA) testing is controversial due to risks of overdiagnosis and overtreatment. Clinical guidelines recommend shared decision-making (SDM) to help patients make informed decisions about screening, yet SDM remains underused in primary care, with documentation rates as low as 10%.

WHAT THIS STUDY ADDS

• This study demonstrates that implementing templated electronic health record (EHR) notes with SDM dot phrases, combined with targeted education, significantly improves SDM documentation rates for PSA screening. The approach is broadly applicable across various clinical areas where SDM is important and can be adopted by all healthcare providers.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The intervention provides a scalable model for integrating SDM into routine primary care, supporting guideline-based cancer screening. Future adoption of similar EHR tools may enhance patient-centred care and standardise preventive health discussions across diverse clinical settings.

Background

Prostate cancer is the second most frequently diagnosed cancer and a leading cause of cancer mortality among men in the USA. The American Cancer Society estimates that in 2025, 313 780 individuals will be diagnosed with prostate cancer, and there will be approximately 35 770 deaths from prostate cancer.¹

Prostate-specific antigen (PSA) testing became clinically valuable for prostate cancer screening in the late 1980s. The Food and Drug Administration then approved the PSA test for monitoring prostate cancer progression in 1986.² Subsequently, it was approved in conjunction with digital rectal examinations for screening asymptomatic men for prostate cancer in 1994. The widespread use of PSA testing in the 1990s led to an increase in prostate cancer diagnoses, and initially, PSA testing was recommended for all men aged 50 years and older,³ predicated on the finding that early detection of clinically significant prostate cancer can increase cure rates. However, following widespread adoption, studies began to show that PSA may not be as effective at screening as initially thought, and there were demonstrated harms associated with overdiagnosis.²⁴,⁶

Many men found themselves subjected to biopsy, treatment and side effects for slow-growing or indolent forms of prostate cancer that likely would not contribute to mortality. Other potential harms associated with PSA testing include anxiety and stress surrounding the testing process, false positives, which are common in men with benign prostatic hyperplasia and lead to unnecessary procedures, and false negatives, especially those associated with aggressive, non-PSA-producing tumours, leading to confusion for patients who are later diagnosed with prostate cancer.^{7 8} This newfound nuance to testing led to changes in guideline recommendations and the endorsement of shared decision-making (SDM) regarding testing to ensure patients’ values and preferences were discussed and respected.^{9 10}

SDM is a collaborative process in which clinicians and patients partner to make healthcare decisions, integrating the best available scientific evidence with patient’s values, goals and preferences. Conversations should be comprehensive of benefits, risks and alternatives, with time allocation determined by the complexity of the decision and patient needs rather than a fixed duration. Clinical guidelines from organisations such as the US Preventive Services Task Force (USPSTF) and the American Urological Association endorse this practice, emphasising the need for clinicians to engage patients in meaningful conversations about screening options.³⁹,¹¹ In its 2018 recommendations, the USPSTF highlights that men aged 55–69 years should make individual decisions about PSA testing after discussion with their healthcare provider, while discouraging routine screening for men aged 70 years and older due to potential harms outweighing benefits.^{12 13}

Despite multiorganisational support for SDM, studies have consistently demonstrated significant underutilisation of SDM conversations in clinical practice. A survey of primary care providers (PCPs) revealed that although most clinicians recognise the importance of SDM for PSA testing, the documentation of these conversations remains low. Data analysed from the National Cancer Institute’s Health Information National Trend Survey 2011–2012 showed that 55% of men between 50 and 74 years have ever had a PSA test.¹⁴ However, only 10% of men were presented with all three components of information required for SDM, regardless of whether they were screened.¹⁵

Understanding the extent to which men are engaged in SDM is essential to gauge how primary care physicians uphold the pillars of medicine, including nonmaleficence, autonomy and beneficence. This quality improvement project aimed to address the gap in SDM documentation for PSA testing within a primary care setting at a safety-net urban resident clinic by implementing structured interventions, including clinician education and standardised templates. The overarching aim was to facilitate consistent and meaningful SDM conversations around PSA screening, improve documentation practices and embed these behaviours into routine clinical workflow. We sought to align screening decisions with patients’ informed preferences, improve the overall quality of care, minimise potential harm from unnecessary screening and empower patients to make decisions that reflect their values. While we did not define a numeric target, this project aimed to establish a sustainable process for improving SDM documentation over time.

Methods

Patient and public involvement

Patients and the public were not directly involved in this study’s design, conduct, outcome selection or dissemination. While patient input was not solicited, the project aims to enhance patient–provider communication.

Setting and baseline data review

SDM for PSA testing is known to be underused nationwide. An internal review within our resident-run clinic revealed multiple barriers contributing to poor SDM documentation rates, including limited resident familiarity with SDM best practices, uncertainty around PSA screening guidelines and how to effectively communicate risks and benefits to patients (figure 1). To assess whether this trend applied to our safety net urban resident clinic with a population of 2678 patients, a pool comprised of all eligible men aged 55–69 years was compiled. The eligible population contained 405 men, of whom 200 were selected using computerised randomisation to form a baseline preintervention cohort table 1.

Figure 1. Fishbone diagram identifying barriers to SDM documentation for PSA screening and the impact of the dot phrase intervention. This fishbone (Ishikawa) diagram illustrates key factors contributing to inconsistent documentation of SDM for PSA screening in the resident clinic. The categories include People (clinician knowledge, habits and training), Process (workflow inefficiencies and time constraints), and Tool (EHR limitations). Implementation of a standardised EHR dot phrase served as the targeted intervention to streamline and improve documentation consistency and completeness. SDM, shared decision-making; PSA, prostate-specific antigen; EHR, electronic health record.

Table 1. Cohort demographics.

Ethnicity	Race	Age range										Average age		% patients
		Patients (n)
		Baseline					Postintervention
		55–59	60–64	65–69	70	Total	54	60–64	65–69	70	Total	Baseline	Postintervention	Baseline	Postintervention
Hispanic/other Hispanic/Puerto Rican	Black/African American	0	1	0	0	1	0	2	0	0	2	63.0	62.0	0.5	0.5
	Other/unknown	21	17	27	2	67	1	42	32	3	128	63.0	61.4	33.5	33.7
	White	0	4	2	1	7	0	4	5	0	11	63.9	63.3	3.5	2.9
	Total	21	22	29	3	75	2	48	37	3	141	63.1	62.2	37.5	37.1
Non-Hispanic	American Indian/Alaskan	0	1	0	0	1	0	0	0	0	1	61.0	59.0	0.5	0.3
	Black/African American	24	41	29	3	97	3	77	55	0	195	62.6	61.9	48.5	50.5
	Native Hawaiian	0	0	1	0	1	0	0	1	0	1	66.0	65.0	0.5	0.3
	Other/unknown	1	0	1	0	2	1	1	3	0	7	62.0	61.4	1.0	1.8
	White	6	12	4	0	22	2	18	3	0	31	61.5	60.7	11.0	8.2
	Asian	0	0	0	0	0	0	1	1	0	6	N/A	59.0	0.0	1.6
	Total	31	54	35	3	123	6	97	63	0	238	62.4	61.6	61.5	62.6
Unknown	Other/unknown	1	0	1	0	2	0	0	1	0	1	61.5	65	1.0	0.3
	Total	1	0	1	0	2	0	0	1	0	1	61.5	65	1.0	0.3
Total		53	76	65	6	200	8	145	101	3	380	62.6	61.6

The selected charts were reviewed for age at various points, including at randomisation and at most recent PSA testing, ethnicity, race, personal history of prostate cancer and presence or absence of a comprehensive SDM conversation regarding prostate cancer screening with PSA during any encounter available in the electronic health record (EHR). The oldest record reviewed was from 2011. Additionally, whether the provider engaging in the SDM conversation was a resident, primary care or a specialist was recorded. Encounters that did and did not qualify as containing SDM were further analysed for common themes and phrasing to form the basis for our intervention dot phrases, an EMR smart phrase that can be quickly populated into a note and modified to enhance documentation and education.

Examples of encounter phrasing that did not qualify as SDM included encounters which simply documented the test as ordered, used phrasing such as “…should have PSA assessment… repeat PSA”, used paternalistic language such as “…will also undergo a PSA test for prostate cancer screening” or justified the need for PSA testing with concern such as low back pain or imaging findings such as nodules, without documentation of a risk and benefit discussion. Again, encounters that included language such as “…patient agrees to PSA testing” or “…patient requested PSA” without evidence of a risk and benefit discussion surrounding the order were not considered qualified SDM.

Examples of encounter phrasing that did qualify as SDM included clear documentation that such a discussion was held, as simply as “…per shared decision making, patient would like to be tested for PSA and further evaluation of prostate cancer if elevated”. Several encounters included extended phrasing that highlighted the risks and benefits of PSA testing, including the possibility of false positives and negatives.

Consistent with national averages, 31.5% of our baseline population had an up-to-date PSA screening on file, completed within the last 4 years, yet only 7.1% had documentation of an SDM conversation with nearly absent follow-up documentation. This preliminary data highlighted a significant gap in SDM documentation and follow-up for PSA testing, underscoring the need for intervention to improve both processes.

Study design

The intervention was informed by the plan–do–study–act (PDSA) methodology, including structured planning, implementation and outcome evaluation; however, it was not structured as a full iterative PDSA cycle. Instead, the project followed a preintervention/postintervention design to assess the impact of our educational and documentation-based strategies. The interventions were implemented over 6 months to assess integration into the clinic’s workflow. After the interventions were implemented, chart reviews were conducted to assess SDM documentation, follow-up documentation and PSA screening rates.

The intervention aimed to improve SDM for PSA screening by incorporating two key components: (1) internal medicine resident PCP education and (2) the use of dot phrases in the EHR system.

The first component involved sending an informational email (online supplemental file 1) to all clinic residents and a brief in-person educational session. This session provided background information on PSA screening, the importance of SDM and guidance on facilitating SDM conversations with patients. In addition, a PSA screening prompt with a hard stop was integrated into the clinic’s standardised EHR template, to prompt providers to assess patient eligibility. This hard stop was followed by the eligibility dot phrase, which generates a series of dropdowns within the EHR to facilitate documentation of the SDM conversation and its outcome, when applicable (figure 2A).

Figure 2. (A) The figure above demonstrates a standardised dot phrase developed to guide and document SDM for PSA screening in men aged 55–69 years. The template begins with a forcing function ‘***’ that requires providers to pause and assess screening eligibility before proceeding, ensuring deliberate and informed engagement in the SDM process. The tool automatically retrieves the most recent PSA result and date (if available) to support informed discussions. It then walks providers through a structured documentation process, including PSA history (e.g., presence or absence of a recent result), patient decision (dropdown menu with options such as ‘declined PSA screening at this time’) and next steps (e.g., ‘order PSA level’ or ‘will revisit again at future appointments’). Each section uses quick-select dropdowns to streamline workflow while maintaining comprehensive documentation. The format promotes efficiency, consistency and adherence to best practices for prostate cancer screening discussions. (B) Demonstrates a dot phrase template for PSA result interpretation and follow-up used to document PSA screening results and guide next steps. It prompts assessment of recent PSA history, screening discontinuation criteria (age >70 or life expectancy <10 years) and provides age-based reference ranges. The tool auto-populates the most recent PSA result and applies correction for 5-alpha-reductase inhibitor use. Follow-up recommendations are provided via dropdown based on PSA level. ARI, androgen receptor inhibitor (e.g., 5-alpha reductase inhibitors such as finasteride or dutasteride); SDM, shared decision-making; PSA, prostate-specific antigen; EHR, electronic health record.

The second component introduced a follow-up dot phrase to streamline the documentation of PSA test results. This dot phrase simplifies the documentation of PSA interpretation and follow-up plans and was designed for ease of use with a series of key presses (figure 2B). A reminder note card was placed near each computer with the prompt.

The dot phrases were created to capture the essential components of SDM, including key PSA screening information, guideline-based recommendations and patient preferences. This structure ensures efficiency, consistency and thorough documentation of SDM discussions.

After the intervention, feedback sessions were conducted with residents which focused on gathering common themes related to obstacles encountered and ease of use. This feedback was used to evaluate the acceptability and feasibility of the intervention within the clinic’s workflow. While additional changes may be considered, the current study focused on valuating a single, structured intervention phase rather than iterative cycles.

Patient eligibility

Patients were eligible for SDM if they were male and either unscreened or had not had a PSA in over 4 years, per National Comprehensive Cancer Network (NCCN) and USPSTF guidelines recommending SDM for men aged 55–69 years or at higher risk (e.g., black men or those with a family history). These criteria ensure that SDM discussions focus on patients who benefit most from individualised decision-making.

Measures

The primary outcome was documentation of SDM conversations on PSA screening, with a goal of increasing documentation from 7.1% preintervention to over 50% postintervention. An SDM conversation was considered eligible if it had occurred within 4 years of the patient’s scheduled visit, regardless of their attendance at the visit that occurred during our 6-month review period. Charts were reviewed as in our baseline data review, and conversations documented with providers outside our clinic were also included.

Secondary outcomes were documentation of follow-up conversations for PSA test results and PSA screening rates. A result was eligible if one of our clinic residents ordered it during our intervention. This measure was tracked to ensure that SDM conversations led to appropriate follow-up for patients undergoing PSA screening. The aim was to achieve follow-up documentation for more than 50% of patients. We also collected data on patient age, ethnicity and race for subcategory analysis.

Process measures were the rate of SDM and PSA follow-up dot phrase use and use of continuity clinic templates.

Balancing measures included abnormal PSA result follow-up as well as feedback from resident providers through in-person sessions at the end of our intervention.

Data analysis

Two independent reviewers conducted a structured EHR audit to ensure data completeness and accuracy, cross-checking their work at intervals to ensure consistency in identifying and recording SDM documentation, PSA testing and follow-up.

Additionally, a structured checklist guided the review process, outlining key elements: assessment of PSA eligibility, documentation of SDM conversations, testing decisions and follow-up actions. This ensured a thorough evaluation of each intervention component, enhancing data accuracy and completeness.

Descriptive statistics summarised quantitative preintervention and postintervention data, focusing on documenting SDM conversations and PSA result follow-ups. Χ² tests compared documentation and screening rates before and after the intervention to determine statistical significance and assess the effectiveness of the intervention.

Resident feedback was analysed using thematic content analysis to identify recurring barriers and facilitators in integrating SDM. Themes related to workflow challenges, time constraints and experiences with the dot phrase were prioritised to provide insights into the intervention’s impact and identify areas for future improvement.

Results

Of the 200 baseline charts reviewed, 14 patients had a PSA ordered but no available encounter associated with the order, meaning SDM documentation could not be assessed and subsequently were not included in our screening eligible cohort. 170 patients had a screening PSA ordered without a qualified SDM conversation documented (n=48), had no encounter associated with the order (n=14) or had never had a PSA ordered and had never been documented to have declined PSA screening (n=108). 14 encounters contained SDM conversations, 2 of which documented an SDM conversation in which the patient declined testing. One encounter with documented SDM and a screening PSA ordered occurred more than 4 years prior to chart review and was not included in the eligible SDM, for a baseline SDM rate of 7.1%.

We found that 90 patients had a prior PSA on file, 3 of whom had a diagnosis of prostate cancer. Of 90 PSA tests ordered, 59 were ordered as a screening PSA within the last 4 years, for a baseline screening rate of 31.5%. The median number of years between the patient’s most recent PSA and our chart review was 2.2 (range=0.5–13.7), and the median age at which patients had most recently been tested was 60 (range=48–69). Notably, when PSA testing was performed, follow-up documentation was nearly absent. Demographics for our baseline and postintervention cohorts were similar, and screening and SDM characteristics for both cohorts are presented in table 2.

Table 2. SDM and screening characteristics of baseline and postintervention cohorts.

			Baseline cohort	Postintervention cohort
Overall number of charts reviewed			200	389
Median age at PSA testing			60	62
Median years since PSA testing			2.2	1.3
Patients with a diagnosis of prostate cancer			3	8
Patients with a screening PSA	SDM documented	Occurred with resident PCP	9	75
		Occurred with specialist	2	2
	No SDM documented		48	63
	No associated encounter		14	12
	Follow-up documented		0	26
	No follow-up documented		59	17
Patients without a screening PSA	SDM documented	Occurred with resident PCP	2	38
		Occurred with specialist	0	0
	No documented SDM		108	131

In the postintervention cohort, numbers within the table represent those collected during the 6-month cycle; however, note that final SDM rates include conversations which occurred within 4 years of the patient’s first eligible encounter within the 6-month cycle. Patients with a screening PSA on file but no encounter associated with the order were unable to be reviewed for SDM and were excluded from final analyses.

PCP, primary care provider; PSA, prostate-specific antigen; SDM, shared decision-making.

The 6-month postintervention cohort included 389 encounters with men aged 55–69 years. The primary outcome, SDM documentation, demonstrated a significant improvement postintervention. The rate of SDM documentation increased from 7.1% preintervention (95% CI: 3.54% to 10.66%) to 37.2% postintervention (95% CI: 32.40%, 42.00%) (p<0.001, χ²=59.23). Although SDM documentation improved significantly, it did not meet our target of 50% of eligible patients. Figure 3 illustrates a statistical process control (SPC) probability analysis of our preintervention and postintervention cohorts, which demonstrated that preintervention, the process we intended to intervene on, completion of SDM in screening eligible men, was stable. Eligible men were stratified by eligibility year in the preintervention analysis using the date of their last PSA screening if there was one on file. If there was no PSA on file, men were stratified using an extrapolated screening due date that assumed screening began at age 55 years and was completed every 4 years until age 69 years. We looked at SDM completion rates over the 6 years prior to our postintervention chart review. Our lower control limit for our preintervention cohort was 0, and the upper control limit (UCL) was 22.8%. As reported above, we see a rise in average rate (p-bar) postintervention. Averages differ slightly from those reported above due to stratification by year and month in the SPC analysis, which excluded a small number of men in each cohort. The rise in SDM documentation rate in screening eligible men to 57.9% in our postintervention SPC analysis, above the UCL of 47.8%, indicates a transition from a stable preintervention process to a changing postintervention process and further supports our intervention as a mechanism of significant change within the system.

Figure 3. Statistical process control p-chart analysis for baseline (preintervention) and postintervention cohorts. Shared decision-making (SDM) completion rate, or the rate of men eligible for prostate-specific antigen screening who had a documented SDM conversation, is on the y-axis and time (years for preintervention and months for postintervention) on the x-axis. P is the SDM completion rate trend line, p-bar is the average of all rates, LCL is the lower control limit (3 SDs below the average with a minimum of 0) and UCL is the upper control limit (3 SDs above the average).

Of those conversations that took place with our clinic residents, our SDM documentation rate improved from 6.0% preintervention to 34.3% postintervention, with 78.7%, or 59, of those conversations being documented using our EHR dot phrase. Despite this improvement, four PSA tests were still ordered without accompanying SDM documentation. Of the patients offered PSA testing during a documented SDM discussion, 49.3% declined. This represents a large increase from our baseline rate of SDM documentation during encounters where a PSA was not ordered of just 1.1% of all eligible encounters or 2 encounters, to 15% postintervention or 38 encounters. 76.3% of those encounters used our EHR dot phrase, similar to our overall rate. We found that 162 screening conversations with eligible patients were missed over the 6-month period.

Our secondary outcome measure focused on documenting appropriate follow-up for PSA testing. Of the 43 patients with a PSA test ordered during our 6-month cycle, 60.4% of encounters included documented follow-up on the result in a subsequent encounter during the cycle, meeting our goal of 50%. Additionally, 7 PSA tests that were ordered prior to our 6-month cycle had documented follow-up during the cycle. We did not track if patients had had a subsequent encounter after the 6-month cycle, and if a patient had not been seen after the test was ordered, they were recorded as not having follow-up. Our interpretation dot phrase was used nine times. Of the screening tests ordered, four were found to be newly abnormal and each abnormal result was appropriately followed up.

PSA screening rates increased, rising from 31.5% preintervention (95% CI: 25.6% to 37.4%) to 37.8% postintervention (95% CI: 33.0% to 42.6%) (p=0.155, χ²=2.02). 25 of those patients screened had never been screened before. While this increase was not statistically significant, the data suggest a positive trend toward higher screening rates. Among patients who had ever been screened, 68.5% were up to date with their PSA testing postintervention, while 40 patients were overdue, 8 of whom declined to retest during a documented SDM conversation.

Contextual elements influencing the intervention emerged during feedback sessions with resident physicians. Residents noted that time constraints and the high acuity of patients with multiple uncontrolled chronic conditions made it challenging to hold SDM conversations consistently. Furthermore, the safety-net nature of the clinic, with its higher disease burden and missed appointment rates, added additional strain on the intervention’s effectiveness. Specifically, of 389 patients who met the criteria and were scheduled for appointments during our intervention, 85 were recorded as ‘no-shows’, resulting in missed opportunities for SDM conversations and PSA screenings.

The intervention’s unintended consequences included an increased administrative burden on residents. Although the dot phrase streamlined SDM documentation, incorporating SDM conversations into busy clinic schedules—particularly with high-acuity patients—proved challenging. This occasionally led to incomplete documentation, with some SDM conversations possibly occurring but not being recorded, likely underestimating the actual SDM rate.

Discussion

Summary

This project aimed to enhance SDM for PSA screening in primary care, focusing on improving documentation and screening rates. The intervention, which included resident education and the implementation of a structured EHR template with a dot phrase, led to a significant increase in SDM documentation—from 7.1% preintervention to 37.2% postintervention. Further, our secondary outcome goal of greater than 50% for follow-up documentation was met, highlighting that attention to SDM documentation and education also led to increased attention to follow-up documentation, although our interpretation dot phrase was used infrequently. This may be partially attributed to missed follow-up appointments or residents forgetting to use the dot phrase—an issue exacerbated by its need for active integration into personalised templates.

The strength of this project lies in its multifaceted approach, combining clinician education with structural modifications to the EHR. This dual approach ensured that SDM was taught and seamlessly integrated into residents' workflows through the dot phrase, enhancing documentation efficiency and standardisation; furthermore, by targeting PSA screening, a key preventive care intervention, the project allowed for measurable improvements in documentation and patient care.

Interpretation

The intervention notably improved the consistency and accuracy of SDM documentation, with a significant statistical increase from 7.1% to 37.2%. This improvement highlights the effectiveness of the structured educational component and EHR modification in addressing gaps in clinical practice. The statistical significance of this improvement demonstrates that the intervention successfully enhanced the SDM process within clinical workflows.

The increase in PSA screening rates, while positive, was not statistically significant. The observed improvement from 31.5% to 37.8% represents a substantial process improvement, even without statistical significance. This 6.3 percentage point increase suggests that the intervention likely positively affected resident behaviour, indicating a trend toward improved SDM discussions. It is important to note that using screening rates alone as a metric carries nuance. As our data revealed, men who may have previously been screened may not wish to undergo screening again following an SDM conversation.

The modest improvement may reflect the complexities of behaviour change, which often requires long-term, sustained interventions and is influenced by various external factors such as patient preferences, healthcare system constraints and competing clinical priorities. Alternatively, the trend may also suggest that our residents previously engaged in SDM conversations at a high rate that was not documented, leading to an increased documentation rate following the intervention but not a significant change in actual screening rates.

Similar studies have shown that while educational interventions can improve SDM documentation, their effects on clinical outcomes, such as screening rates, are more variable. For instance, Kinnersley et al¹⁶ found that training interventions to improve SDM documentation led to better recording of SDM discussions but did not necessarily result in improved patient outcomes, such as adherence to screening guidelines.¹⁶ Similarly, Elwyn et al¹⁷ found that while structured educational initiatives improved documentation of SDM conversations, the impact on clinical behaviours, including screening rates, was inconsistent.¹⁷ These findings align with the results of this study, which showed improvements in documentation but only modest increases in screening rates, underscoring the challenges of translating improvements in documentation into measurable clinical outcomes.

The intervention had a positive impact on both residents and patients. For residents, it offered an opportunity to improve their understanding and practice of SDM, enhancing their ability to document SDM conversations more effectively. This improvement in documentation skills will likely lead to better patient–provider communication and more informed decision-making. For patients, the increase in SDM documentation likely facilitated more transparent and thorough discussions regarding PSA screening, which could result in more informed and thoughtful decisions. On a systems level, the EHR modification enabled standardised and easily accessible documentation, promoting better follow-up and continuity of care.

While the intervention improved SDM documentation, the increase in PSA screening rates did not reach statistical significance. Several contextual factors likely influenced this outcome. The complexity of behaviour change in clinical practice and challenges such as time constraints, high patient volume and competing priorities likely contributed. Additionally, variations in resident engagement and institutional culture could have affected the consistency of SDM implementation. External factors, such as patient preferences and their decision-making processes, also play a critical role in screening decisions, and these may not always align with clinical recommendations. These contextual barriers are common in clinical practice and highlight the multifaceted challenges of implementing SDM interventions.

The intervention required minimal direct financial investment, but there were notable opportunity costs associated with the time faculty and residents spent engaging with the project. Despite the modest increase in screening rates, the improvements in SDM documentation suggest that future efforts could refine the intervention to address additional challenges. These could include enhancing patient follow-up and optimising clinical workflows. A more comprehensive approach—integrating streamlined systems and additional strategies for engaging patients—may yield more pronounced clinical outcomes and justify the time investment and opportunity costs involved.

Limitations

This study was conducted at a single safety-net resident clinic in an urban setting, which may limit the generalisability of the findings to other institutions with different patient populations or healthcare systems. The patient demographics and health conditions may vary at other institutions, potentially affecting the applicability of the results.

We did not incorporate pre-education or post-education tests or surveys into our study to evaluate resident understanding and learning. Although we are making the assumption that every resident was equally effective and willing to have an SDM conversation with patients, it’s likely that resident understanding and explanation varied, which may have also affected patient’s decisions to screen or decline, or a resident’s decision to use SDM when seeing patients.

Furthermore, the study’s retrospective nature introduces the possibility of incomplete or inconsistent data in the EHR and variability in clinician reporting practices. Confounding factors, such as differences in patient behaviour or unmeasured variables, could not be fully controlled, which may have influenced the outcomes.

To address data gaps, rigorous data cleaning and verification processes were employed to minimise bias and improve internal validity. Future prospective studies could mitigate these limitations and provide further insights into the intervention’s effectiveness.

Conclusions

This project demonstrates the effectiveness of a structured educational intervention combined with EHR modifications to improve SDM documentation and PSA screening rates. It offers a practical approach to enhancing preventive care by improving communication between providers and patients. The sustainability of this intervention is promising, particularly due to the EHR modifications that can be maintained over time. However, ongoing support and periodic reinforcement of SDM training will be necessary to sustain and possibly further improve outcomes. Ensuring residents are trained on the intervention is crucial for its long-term success.

The findings suggest that similar interventions, especially EHR systems, could be implemented in other primary care settings. Although this intervention was implemented in a safety-net clinic, SDM and EHR integration principles can be applied across various healthcare environments. Future studies could assess the adaptability and scalability of this intervention in different clinical settings.

This project has significant implications for integrating SDM into routine care processes and addressing educational and structural barriers. Future research could explore optimising SDM in other areas of preventive care, enhancing patient engagement in decision-making and evaluating the long-term impact of SDM interventions on clinical outcomes. Expanding the intervention to include attending physicians and other healthcare providers will be important for its scalability. Additionally, incorporating patient preferences more explicitly into SDM and providing follow-up support and reminders may improve screening rates and patient outcomes.

Data availability statement

Data are available on reasonable request.