GME Patient Safety Quality and Innovation Symposium

Issues Addressed/Background

Patients with suspected malignancy are routinely admitted inpatient for diagnostic evaluation for urgent but not emergent workups, utilizing limited inpatient resources. We wondered if some of these workups could be managed outpatient in a timely manner with the right infrastructure. Conventional oncology referral pathways follow a consultation-first sequence; ie patients are seen, then tested, then redirected, which delays diagnosis and consumes limited clinic capacity. The UCSD Suspicion of Cancer (SoC) clinic was designed around an inverted model: physician-led triage occurs right at referral, testing is coordinated before the first visit, and patients who can be safely redirected to disease-specific teams never need an SoC appointment. This report describes operational outcomes over the first four months following clinic launch in December 2025.

Description of the Project

The SoC clinic serves patients with findings suspicious for malignancy without a confirmed site-specific diagnosis. Referrals are restricted to ED providers, inpatient teams, and specialty oncology clinics. The referral order set was kept deliberately simple: two questions asking whether findings are suspicious for cancer and whether a known primary exists, to decrease friction with referrals. Core operational components included mandatory CT chest/abdomen/pelvis for all ED-placed referrals, a target of first patient contact within 1–2 business days, an expedited outpatient biopsy and imaging pathway, and a new patient access supervisor with expertise in scheduling, insurance verification, and records retrieval. Clinic capacity was set at approximately 2 SoC slots per week for now during the initial phase.

Lessons Learned

Between December 2025 and March 2026, 65 patients were referred to the SoC clinic. Of these, 30 (46%) were diverted to disease-specific oncology teams or primary care without an SoC visit, and 16 (25%) completed SoC appointments. The remainder involved insurance denials (n=5), patient refusals or no-shows (n=5), and cases pending or closed for other reasons. Cases were diverted from in-person SoC clinic (to subspecialty oncology clinics) following triaging actions, including the following: ordering additional imaging or other testing, placing IR biopsy referrals via expedited pathways, correcting misdirected referrals, and communicating this directly with patient families. This allowed for SoC staff to contact patients within a few days to begin the urgent outpatient workup and place referrals soon after, thereby decreasing time to workup compared to having to schedule clinic visits and have a formal visit to begin evaluation. 

Recommendations/Next Steps

The most meaningful structural improvement is establishing disease-team liaisons (designated oncologists on each disease team) who are comfortable accepting early referrals based on triage data alone, so patients can bypass the SoC visit when workup already points toward a specific diagnosis. Addressing patient follow-through will require a more active outreach protocol after initial contact, particularly for pre-visit orders. Future data collection should include time from referral to imaging completion, time to diagnosis, and longitudinal tracking of whether the 2-slot-per-week capacity remains appropriate as referral volume grows. Additionally, we would like to eventually open up the clinic referrals to be placed by PCPs and outpatient providers.

Background: Induction of fetal demise is an advanced procedural skill in Obstetrics and Gynecology (OBGYN) that expands patient access to later-gestation abortion care while mitigating risk of unintended live periviable births. Additionally, the literature shows that many patients prefer induction of fetal demise prior to termination of pregnancy when presented with the option. Despite this, ultrasound-guided injection for fetal asystole is an under-taught skill in abortion training programs, most OBGYN residencies, and many Complex Family Planning (CFP) fellowships. A simulation for this procedure could be a steppingstone for clinicians to learn this skill in a low-risk setting before incorporating it into clinical practice.

Description: This project aims to increase the average clinician comfort and confidence score when performing ultrasound-guided needle procedures such as induction of fetal asystole following standardized simulation training among the participating faculty clinicians. Clinician participants included attendings and fellows from the CFP and the Maternal-Fetal Medicine (MFM) divisions. Study team members used the Rubin Model to conduct standardized simulation of injection for fetal asystole (Figure 1). The Rubin Model is a high-fidelity system consisting of three nested ballons (one which represents the fetal heart, one which represents the fetal body, and one which represents the amniotic sac) specifically designed to teach ultrasound-guided intra-amniotic, intrafetal, and intracardiac injections. Other educational materials developed for this project include a division-wide protocol to facilitate introduction of this skill into clinical practice and a patient-facing Frequently Asked Questions document to address common questions and concerns that patients may have when being counseled on the risk/benefits/alternatives of undergoing injection for fetal asystole. These documents were reviewed with participants prior to procedural simulation along with relevant evidence on diVerent pharmacologic agent options and diVerent ultrasound techniques (Figure 2). Clinician comfort and confidence was assessed using pre-simulation and post-simulation 10-question 5-point Likert scale surveys. Questions covered areas including general confidence, technique and equipment, patient safety and communication, and overall preparedness.

Lessons learned or expected outcomes: Seven faculty clinicians participated in the standardized simulation of injection for fetal asystole and took the pre- and post-surveys. With success defined an average post-training score of 4.0 or higher across all ten questions OR an absolute increase of 1.0 points from the pre-training average score, this simulation had a 72% success rate. Average post-training scores increased by 36% (or 1.2 points out of 5) from average pre-training scores (Figure 3).

Recommendations/next steps: UCSD CFP faculty are now trained for and will begin to oVer injection for fetal asystole to appropriate patient candidates. This simulation will be conducted for CFP and MFM fellows on a yearly basis. Ultimately, the goal is to help clinicians expand their skillsets to include this procedure, thereby improving patient access to second-trimester abortion.

Background

The Emergency Department (ED) is often a place of high-stress or difficult cases, with very little time to process these events before returning to other duties in the department. “Hot debriefs”, or short discussions immediately following a difficult case, such as cardiac arrest, pediatric case, or high-stress situation have been shown to decrease burnout, reduce stress, and lead to key process improvements relevant to future codes. Many different styles of hot debriefs have been described, including the Take STOCK, BONE break, and TIM tool. Previous studies have compared hot vs cold debriefs, or examined standardized debrief implementation in the prehospital or ICU setting. In this project, we sought to implement the STOP 5 debrief tool immediately following cardiac arrests, pediatric codes, and other difficult cases in the ED, address physicians perceived lack of time to debrief, and improve confidence of those leading debriefs.

Methods

Nursing staff and residents were educated on the STOP 5 debrief during shift change and during educational conferences. Laminated posters were placed in the resuscitation rooms and in the “doc boxes”, areas of the ED most often occupied by attendings and residents. Standardized surveys were distributed to attendings and residents (PGY1-4) before implementation of the standardized tool and again 3 months later to gauge impressions and attitudes towards the STOP 5 debrief. Continuous variables were calculated with independent samples T-test, ordinal responses were calculated with Mann-Whitney U test, and binary responses were calculated with Fischer's exact test.

Results

A total of 50 participants completed pre-intervention surveys, and 26 participants completed post-intervention surveys. On a rating scale of 1-5, there was a significant increase in the number of physicians who felt comfortable initiating a debrief (3.28 vs 4.00, p-value = 0.036). On a rating scale of 1-5, there was no significant change in the level of confidence with regards to leading a successful debrief (3.2 vs 3.7, p-value = 0.169). The most commonly cited reasons for not having a debrief include lack of time (84%), lack of experience (30%), and “not my role” (15%), however, participants almost universally wanted more debriefs.

Conclusions

The implementation of a standardized hot debrief led to an improvement in how comfortable physicians felt initiating debriefs. There remains significant barriers to universal debriefs after appropriate cases, most notably physicians perceived lack of time and lack of confidence in leading these debriefs. Further work is needed to build a culture where hot debriefs are expected and encouraged , while further training is also needed to help ED staff feel more comfortable leading these debriefs.

Background

Access to emergency contraception (EC) is a key component of reproductive healthcare and time-sensitive pregnancy prevention. Two commonly used EC formulations include ulipristal acetate, a prescription medication that remains effective for up to 5 days after unprotected intercourse and across a wider body mass index (BMI) range, and levonorgestrel-based EC, an over-the-counter medication that is effective at pregnancy prevention for up to 3 days after unprotected intercourse. Clinical guidelines recommend ulipristal acetate as first-line EC, particularly for individuals with higher BMI or longer delay after unprotected intercourse. However, pharmacy stocking practices may limit timely access to ulipristal acetate, potentially creating disparities for patients with higher BMI and those facing structural barriers to care. This quality improvement project sought to evaluate the availability and dispensing timelines for stocking EC across San Diego pharmacies.

Description

Study team members conducted standardized telephone calls to retail pharmacies across San Diego using a structured script (Figure 1). Pharmacies were asked whether they stock ulipristal acetate for same-day prescription pickup, and the expected timeline for availability if it needed to be ordered. Availability of levonorgestrel EC and same-day access were also assessed. Responses were recorded and analyzed using descriptive statistics.

Lessons learned or expected outcomes

A total of 129 of 132 (97.7%) San Diego retail pharmacies were able to be contacted. Levonorgestrel EC was immediately available at 82 pharmacies (63.6%), compared to ulipristal acetate at 39 pharmacies (30.2%).

Thirty-three pharmacies (25.6%) stocked both ulipristal acetate and levonorgestrel EC, while 49 pharmacies (38.0%) stocked levonorgestrel EC but not ulipristal acetate. Additionally, 41 pharmacies (31.8%) stocked neither product at the time of contact (Figure 2).

Among pharmacies that stocked ulipristal acetate EC, the average number of tablets currently stocked was 1.8 (± 1.5 SD) with a minimum of 1 and a maximum of 9 tablets available. For the 85 pharmacies where ulipristal acetate was not currently stocked, and information was available, 2 pharmacies (2.4%) reported access to ulipristal acetate the same day, 50 pharmacies (85.5%) reported availability within 24 hours, and 20 pharmacies (23.5%) reported availability within 48 hours through distributor ordering or transfer from nearby locations.

Recommendations/Next Steps

Levonorgestrel EC is widely accessible across San Diego pharmacies, whereas ulipristal acetate availability remains limited and often dependent on ordering delays. Over 63% of pharmacies stocked levonorgestrel EC while only 30% stocked ulipristal acetate, highlighting a substantial access gap between EC formulations. Even when pharmacies can obtain ulipristal acetate within 24–48 hours, delays reduce the effectiveness of this time-sensitive medication. Limited on-site inventory further constrains immediate dispensing capacity and may disproportionately affect patients with higher BMI.

Ongoing quality improvement efforts include provider education, development of a clinician-facing EC pharmacy directory, and repeat pharmacy assessments to determine whether increased prescribing demand improves pharmacy stocking and timely access to ulipristal acetate.

Background:  

Due to uneven geographic distribution of health care infrastructure, long distance travel is a reality of cancer care for many patients. Patients travel significant distances to access tertiary academic centers or receive multidisciplinary care¹. Outcomes for patients have been mixed. A 2021 study of HCC patients undergoing liver resection showed living >50 miles from high volume tertiary academic centers was associated with worse overall survival².

However, a subsequent NationalCancer Database (NCD) study demonstrated that HCC patients who travel >30 km/18.6 mi were more likely to have higher rates of locoregional/surgical treatment, academic center care, and improved overall survival. Authors reflected this population may be highly motivated and financially able to seek distance care³. Further outcomes research is needed, especially for HCC patients receiving longitudinal multidisciplinary therapy.  

UC San Diego (UCSD) services a large catchment area for cancer patients. Our goal is to better understand the demographics of patients with HCC at UCSD, identify protective factors that enable patients to travel long distances, and use this information to tailor care. 

Description:

We are conducting a retrospective review of 381 patients with HCC who received care at UCSD, stratified by travel distance. Distance travelled was measured by Google Maps from treating facility to home zip code.  Similar to prior studies^2,4, distance was stratified at 50 miles and 100 miles. Correlation is planned with patient, disease, and treatment-specific factors, as well as primary outcomes of clinical trial enrollment and overall survival. 

Lessons Learned or Expected Outcomes: 

Of the 381 patients identified with HCC, 265 received systemic therapy at UCSD. Travel ranged from 1.9 to 308 miles for care. Excluding 4 outlier patients who traveled 164-308 miles, the remaining 261 patients are displayed in the figure below. 198 patients, (74.7%) live within 50 miles, 25 patients (9.4%) live between 50-100 miles, and 42 patients (15.8%) live over 100 miles away. 

Additional data analysis is underway.  

Roughly a quarter (25.2%) of patients with HCC travel over 50 miles to receive care at UCSD. Understanding the factors that make this possible for patients can guide institutional support programs and outreach.  

Recommendations:  

Next steps include completing our retrospective patient chart review. If factors like caregiver presence, language interpreter availability, or household support are protective for distance travel, these may help guide support programs.   

Background: Advance care planning is essential to providing medical care that is aligned with patients’ goals and values. However, conversations regarding goals of care and life-sustaining interventions are often initiated late in the course of illness, such as during hospitalization for critical illness or after loss of decision-making capacity. For high-risk patients in particular, proactive initiation of these discussions in the primary care setting is important to ensure patients’ preferences are known and respected throughout their medical care.

As part of a nationwide quality improvement initiative at the VA, the Life Sustaining Treatment (LST) note was implemented to document patients’ preferences regarding resuscitation and other life-sustaining interventions. LST documentation is especially important for high-risk patients who are at an increased risk for hospitalization and mortality—defined as having a Care Assessment Needs (CAN) score of ≥ 95. The rate of LST note completion for high-risk patients at the San Diego VA is 27%, compared to the national VA average of 49%. To address this need, we conducted a quality improvement initiative to increase advance care planning discussions and LST documentation for high-risk patients in the VA primary care setting.

Methods: Patients with a CAN score of ≥ 96 and no prior LST documentation were identified across eight VA primary care resident clinic panels. Dedicated appointments for up to 30 minutes (in-person, video, or phone) were scheduled for each patient. Family members were encouraged to join if the patient preferred. Educational materials with detailed information regarding advance care planning and life-sustaining treatments were mailed to each patient for review 1-2 weeks prior to the appointment. One medical provider conducted all appointments. Patients' health care goals and life-sustaining treatment preferences were discussed, and an LST note was completed along with an advance directive (if not previously completed) and POLST form (if applicable). Resuscitation status orders were updated in the VA Computerized Patient Record System (CPRS).

Results: Of the 10 patients identified as high-risk, 8 patients agreed to participate in this intervention. 5 patients (63%) had not previously discussed advance care planning with a medical provider. 2 patients (25%) had not previously completed an advance directive. 4 of 5 patients (80%) with a do-not-resuscitate status did not have a previously completed POLST form. During the 4-week intervention period, there was a 100% rate of LST note, advance directive, and POLST form completion.

Conclusions: We have demonstrated an effective approach to integrating advance care planning and LST documentation for high-risk patients in the primary care setting. An important next step involves developing a team-based workflow to routinely access the CAN score database, identify patients with high CAN scores, and notify providers. In addition, surveying patients about their experience participating in this intervention can help assess the impact and benefit of these discussions and identify areas for improvement. Future efforts will also focus on increasing provider education and engagement in this initiative across other VA primary care clinics.

INTRODUCTION: Frail patients are at significantly elevated risk of major postoperative complications and mortality. The Risk Analysis Index (RAI) Frailty Index is a validated tool for preoperative frailty assessment. In the Veterans Affairs (VA) system, the Surgical Pause Practice is requiring preoperative frailty index documentation. However, as of September 1, 2025, no frailty scores were being recorded on general surgery cases at our VA. Because the preoperative clinic represents the most consistent and controllable point of contact for surgical patients, this project focused on clinic-based workflow integration to maximize impact. This project aimed to increase documentation of preoperative RAI frailty scores among general surgery patients from 0% to 25% over a 4 month time period. METHODS: This quality improvement project used the Institute for Healthcare Improvement’s Model for Improvement. An interprofessional team of Chief Residents of Quality Improvement and Safety, licensed vocational nurses (LVNs), and general surgery residents and faculty was formed. Process mapping and stakeholder interviews identified that lack of provider knowledge about frailty and no incorporated frailty evaluation in clinic workflow contributed to an absence of frailty data. To address these barriers, providers were first taught about the frailty index during our bi-weekly conference. Then, a standardized workflow was implemented to integrate frailty assessment into routine clinic operation. A patient-facing questionnaire derived from the RAI frailty index and a note template in the Computerized Patient Record System (CPRS) were developed. LVNs administered the questionnaire to all eligible patients during intake. Eligible patients were those being evaluated for potential surgery. Excluded patients were those presenting for post-operative follow-up or being managed nonoperatively. Providers then conducted their usual clinical evaluation. If a patient was deemed appropriate for surgery, providers used the completed questionnaire to input a frailty index note in CPRS. Iterative Plan-Do-Study-Act (PDSA) cycles were conducted to refine implementation. The primary outcome was the percentage of general surgery cases with a documented frailty index. The process measure was questionnaire completion. A balancing measure of clinic workflow impact was assessed qualitatively. RESULTS: From February 1, 2026 to February 28, 2026, 63 eligible patients were identified and booked for surgery. 20 patients had frailty indices recorded during this time frame, representing a 31% increase from baseline. Questionnaire completion reached 100% of eligible clinic patients. Two qualitative balancing measures were identified. First, the availability of clinic supplies initially impaired questionnaire completion, rectified on subsequent PDSA cycles with more supplies. Second, turnover among trainees led to inconsistent access to the RAI calculator, intermittently limiting completion of frailty documentation. CONCLUSION: A patient-facing questionnaire plus clinician education and workflow integration can lead to improved frailty screening for general surgery patients. Embedding the questionnaire within existing team roles, specifically LVN clinic intake, was critical to success. Limitations of this study include single-clinic implementation, short follow-up duration, and reliance on rotating trainee teams, which may affect sustainability and reproducibility. Future work will focus on standardizing access to the RAI frailty calculator and expanding implementation to other surgical clinics and non-clinic areas within our hospital.

Issues Addressed / Background 

Informed consent is a foundational component of ethical, patient-centered care, particularly in procedural specialties such as interventional radiology (IR). At its core, consent is not merely a legal requirement, but a critical opportunity to build trust, foster meaningful patient-clinician partnerships, and support shared decision-making. High-quality consent conversations enable patients to ask questions, express their values and concerns, and make informed, personalized decisions about their care. 

Despite this, most clinicians receive limited, if any, formal training in how to conduct these conversations effectively. In high-throughput procedural environments, consent is often compressed into a time-constrained, transactional process, limiting opportunities for dialogue, understanding, and alignment with patient goals. In IR, where patients may first meet their proceduralist immediately prior to an intervention, these challenges are amplified. This highlights a critical need for structured training that supports relational, communicative, and ethically grounded consent practices. 

Description of the Innovation 

iCONSENT is an ethics- and humanities-based curriculum designed to improve the quality of informed consent conversations across the procedural care team. Developed by the Interventional Initiative, a 501(c)(3) not-for-profit organization, the curriculum reframes consent as a collaborative process centered on patient understanding, values, and trust. 

The iCONSENT curriculum includes six modules and integrates: 

- Evidence-informed communication frameworks to support clear, patient-centered discussions 

- Practical tools for assessing understanding, addressing uncertainty, and guiding shared decision-making 

- Varied and interactive learning methods, including didactics, role-playing, case-based discussions, and animated videos 

- Humanities-based reflective components (e.g., narrative, visual art, and poetry) to explore the emotional and ethical dimensions of patient care 

The curriculum is designed for integration into real-world training and continued education and is accessible both as a self-directed online curriculum and as a live, facilitated training program. An in-person pilot launch of iCONSENT was held at UC San Diego in August 2025 with participation from diagnostic and interventional radiology trainees, medical students, and faculty. 

Lessons Learned / Expected Outcomes 

Early implementation demonstrates strong engagement across learner groups and highlights a previously unmet need for structured training in consent communication. Preliminary evaluation demonstrates improvements in learner-reported confidence and communication approaches, particularly in facilitating more open, patient-centered conversations. 

Anticipated outcomes include: 

- Improved quality and consistency of consent conversations 

- Enhanced patient understanding, trust, and engagement in decision-making 

- Stronger alignment between clinical recommendations and patient goals 

- Increased clinician confidence in navigating complex or emotionally charged discussions 

By emphasizing the relational and ethical dimensions of care, iCONSENT may also help mitigate moral distress and contribute to more meaningful clinical interactions, though this remains an area of ongoing study. 

Recommendations / Next Steps 

iCONSENT represents a scalable approach to strengthening informed consent as a core clinical skill grounded in trust, communication, and partnership. 

Next steps include ongoing assessment of the durability of training effects on clinician communication and shared decision-making skills, as well as evaluation of its impact on clinician well-being, including burnout and moral injury. Expansion efforts are underway to implement iCONSENT across additional institutions and procedural specialties, both nationally and internationally. 

Future directions include formal evaluation of the curriculum’s impact on patient experience, comprehension, and decision-making.