GME Patient Safety Quality and Innovation Symposium

Background: America is in the midst of a syphilis epidemic. The CDC documented 207,255 new cases of syphilis in 2022, representing the highest incidence of disease observed since 1950 and a 17.3% increase from 2021. Treponemal central nervous system (CNS) invasion is estimated to occur in approximately 25-60% of infections, however, the current standard of care for non-neurologic syphilis does not provide antimicrobial dosing sufficient to achieve CNS clearance. This raises concern for a preventable burden of neurologic morbidity secondary to treponemal CNS persistence and subsequent neurosyphilis. Our central hypothesis is that individuals with a history of syphilis infection will experience greater odds of neurologic morbidity relative to their uninfected counterparts.

Methods: A preliminary 10-year retrospective analysis (01/2013-12/31/2022) of deidentified electronic medical records (EMR) from an academic health system was undertaken utilizing Epic SlicerDicer. The exposure of interest consisted of grouped ICD-10 codes representative of syphilis infection, with neurologic outcomes defined as ICD-10 diagnoses reflective of the principle clinical manifestations of neurosyphilis. Exploratory analysis of psychiatric outcomes was performed understanding psychiatry to be a property emergent of neurology. Unadjusted, bivariate analysis was performed for preliminary assessment of exposure-outcome association.

Results: 3,619,941 patient EMR were included in the final analysis. 5,943 patients were found to have a diagnosis of syphilis. Patients with a history of syphilis had a higher prevalence of diabetes mellitus (DM), hyperlipidemia (HLD), and HIV relative to patients without a history of syphilis. Additionally, patients with syphilis exposure were found to have significantly higher odds of neurologic morbidity outcomes including stroke (OR = 5.14; 95%CI = 4.71-5.61), dementia (OR = 20.68 ; 95%CI = 19.27–22.19 ), sensorineural hearing loss (OR= 5.75; 95% CI= 5.07-6.2), and blindness (OR= 15.74 ; 95%CI = 14.31-16.54) relative to unexposed patients. Exposed patients also had significantly higher odds of psychiatric morbidity across all mental health outcomes assessed. Sensitivity analysis was performed whereby bivariate exposure-outcome associations were reassessed within a cohort restricted to patients without the potential confounding diagnoses of DM, HLD, and HIV. All neurologic and psychiatric outcomes studied were observed to retain significance during confounder-restricted sensitivity analysis.

Conclusion: Syphilis infection was associated with significantly increased odds of neurologic and psychiatric morbidity across all outcome measures queried after controlling for several potential confounders. These data are concerning as they suggest the possibility of a clinically significant limitation to the neurologic coverage provided by the current standard of care for non-neurologic syphilis. We plan to more rigorously evaluate our hypothesis in a follow-up study that will utilize multivariate logistic regression analysis to more precisely determine the independent effect of syphilis exposure while adjusting for a number of potential confounders.

Background: America is in the midst of a syphilis epidemic. The CDC documented 207,255 new cases of syphilis in 2022, representing the highest incidence of disease observed since 1950 and a 17.3% increase from 2021. Treponemal central nervous system (CNS) invasion is estimated to occur in approximately 25-60% of infections, however, the current standard of care for non-neurologic syphilis does not provide antimicrobial dosing sufficient to achieve CNS clearance. This raises concern for a preventable burden of neurologic morbidity secondary to treponemal CNS persistence and subsequent neurosyphilis. Our central hypothesis is that individuals with a history of syphilis infection will experience greater odds of neurologic morbidity relative to their uninfected counterparts.

Methods: A preliminary 10-year retrospective analysis (01/2013-12/31/2022) of deidentified electronic medical records (EMR) from an academic health system was undertaken. The exposure of interest consisted of grouped ICD-10 codes representative of syphilis infection, with neurologic outcomes defined as ICD-10 diagnoses reflective of the principle clinical manifestations of neurosyphilis. Exploratory analysis of psychiatric outcomes was performed understanding psychiatry to be a property emergent of neurology. Unadjusted, bivariate analysis was performed for preliminary assessment of exposure-outcome association.

Results: 3,619,941 distinct patient EMR were included in the final analysis. 5,943 patients were found to have a diagnosis of syphilis. Patients with a history of syphilis had a higher prevalence of diabetes mellitus (DM), hyperlipidemia (HLD), and HIV relative to patients without a history of syphilis. Additionally, patients with syphilis exposure were found to have significantly higher odds of neurologic morbidity outcomes including stroke (OR = 5.14; 95%CI = 4.71-5.61), dementia (OR = 20.68 ; 95%CI = 19.27–22.19 ), sensorineural hearing loss (OR= 5.75; 95% CI= 5.07-6.2), and blindness (OR= 15.74 ; 95%CI = 14.31-16.54) relative to unexposed patients. Exposed patients also had significantly higher odds of psychiatric morbidity across all mental health outcomes assessed. Sensitivity analysis was performed whereby bivariate exposure-outcome associations were reassessed within a cohort restricted to patients without the potential confounding diagnoses of DM, HLD, and HIV. All neurologic and psychiatric outcomes studied were observed to retain significance during confounder-restricted sensitivity analysis.

Conclusion: Syphilis infection was associated with significantly increased odds of neurologic and psychiatric morbidity across all outcome measures queried after controlling for several potential confounders. These data are concerning as they suggest the possibility of a clinically significant limitation to the neurologic coverage provided by the current standard of care for non-neurologic syphilis. We plan to more rigorously evaluate our hypothesis in a follow-up study that will utilize multivariate logistic regression analysis to more precisely determine the independent effect of syphilis exposure while adjusting for a number of potential confounders.

Background:

Bridging the communication gap between General Surgery and Anesthesiology residents is critical for enhancing team performance in emergent situations within the perioperative setting. Historically, these residents have practiced simulation training independently, despite the necessity for interprofessional collaboration in their daily practice. This disconnect can lead to compromised patient safety, poor work relations, and physician burnout.

Description of the Project:

To address this, we introduced an interprofessional simulation training program. The program utilizes high-fidelity mannequins for anesthesiology training and box trainers or live pigs for surgical training. Designed to be appropriate for various training years (Table 1), the simulations were written by anesthesia attendings and residents to focus on joint decision-making, enhancing communication, camaraderie, confidence, and respect among participants. General surgery attendings collaborated with the authors to ensure final simulations were of appropriate complexity and included actionable surgical concerns. Funding is supported by a $10,000 Academy of Clinician Scholars grant.

Lessons Learned/Expected Outcomes:

Post simulation surveys include the NASA Task Load Index (TLI) and questions related to perceived team performance. The TLI is used to measure the effectiveness of the purposefully induced stress response. The remaining questions will be used to assess participants’ perception of their communication, camaraderie, confidence, respect, and understanding of the skills of other team members in the perioperative environment. Longitudinal participation is expected to show significant differences where a single interprofessional simulation event may not. The first series of simulations are incomplete with live animal sessions pending. Once completed, TLI values from live pig sessions will be compared to sessions when they were unavailable or inappropriate. Live animal simulations are expected to increase the stress response of all participants.

No statistically significant differences were observed nor expected with current sample size (62 survey respondents, 45 anesthesia, 17 general surgery). However, interim data analysis does suggest emerging trends. Reported mental demand, temporal demand, effort, and frustration on the TLI decreased with seniority (Figure 1). Surgery reported lower mental demand, temporal demand, effort, and frustration but higher performance demand than anesthesia (Figure 2). Average scores within specialties showed: decreased temporal demand with increased anesthesia year; increased physical and performance demand with surgical year. The data revealed trends of increasing measures of perceived team performance with class year in nearly all categories (Figure 3). No overall trends were observed between specialties (Figure 4). However, increasing averages were observed in all categories when comparing within specialties.

Recommendations/Next Steps:

The project will continue over multiple years to assess longitudinal interprofessional simulation training on medical education. Additional specialties, starting with OB/GYN in the ’24-’25 academic year, will be incorporated. Annual interim data analyses will ensure the program's adaptability and effectiveness. Continuous improvement in simulation scenarios and training methods will be prioritized, with the ultimate goal of establishing a best-practice model for interprofessional simulation training in medical education. This model will be made available to simulation programs at other institutions via a Joint Simulation textbook with Springer Nature.

References:

Not applicable. All content is original authorship by the project team.

Background:

 The incidence of induced labor has tripled from 9.5% to 31.4% between 1990 and 2020, due to an increase in the incidence of maternal and fetal indications for induction, as well as an increased recognition of the safety of induction of labor by maternal request at 39 weeks of pregnancy1,2,3. Induced labor has been shown to lead to decreased antepartum office visits and postpartum hospitalizations, with an overall neutral impact on healthcare utilization4,5. However, the increased time and number of interventions on labor and delivery for patients undergoing induction can be challenging for patients and the health care system.

 Induction of labor practices vary among physicians, midwives, and nursing staff, driven by both patient and provider preferences. Furthermore, UCSD cesarean birth rates for induced labor are higher than that for spontaneous labor. Given that induction itself is not thought to increase the risk of cesarean, this may be attributable to these patients’ comorbidities or differences in management practices during induction.

 There is a clear need to understand and increase the use of evidence-based labor induction practices that could decrease time to birth during inductions and decrease the rate of cesarean birth. 

Goals:

-Characterize patient characteristics, induction practices, and decisions leading to cesarean birth in birthing persons undergoing induction of labor

-Improve the implementation of provider and institution best practices for induction of labor and cesarean birth decisions

-Decrease the rate of cesarean birth in induced labor at UCSD

Interventions:

Implement an induction of labor education intervention consisting of evidence-based best practices for induction of labor. Educational talks and a handout on labor and delivery were rolled out in January 2024.

Preliminary Results:​

No apparent increase in cesarian rate since implementation of the labor checklist protocol. No apparent concerns for safety of interventions.

Next Steps:​

Granular data analysis of the impact on labor induction techniques with introduction of the labor checklist from Nov 23 – Feb 24. Analysis will include pre and post-intervention comparisons of:​

NTSV cesarean birth rate; ​Percentage of patients undergoing combination, individual, or staged approaches to cervical ripening​; Percentage of patients undergoing early amniotomy​; Maximum dose of oxytocin used; oxytocin discontinuation during labor​

Expected Outcomes:​

We expect that evidence-based induction of labor practices will increase in frequency in the period following the implementation of the checklist without an adverse impact on the cesarean delivery rate.​

Next steps:​

If evidence-based practices do not improve in frequency as expected, surveys of labor and delivery staff could be undertaken to identify challenges in implementing these practices and appropriate next steps.​

Background: 

Evidence based procalcitonin algorithms recommend discontinuation of antibiotics at values < 0.25 ng/mL for lower respiratory tract infections. A retrospective study conducted at our academic center in 2021 demonstrated that treatment was continued despite a low PCT value in 80.4% of patients on antibiotics. We examined the impact of a nudge intervention on clinicians' decision to de-escalate antibiotics in patients with a low procalcitonin value. 

Methods: 

We conducted a pre-post intervention study on adult inpatients with a procalcitonin result < 0.25 ng/mL receiving antibiotics for LRTIs. We excluded patients on antibiotics for non-LRTIs, COPD exacerbation, or VAP/HAP, as well as immunosuppressed and ICU patients. In the intervention stage, we used a dynamic EPIC procalcitonin report to identify eligible patients and contacted 1st call providers via EPIC Secure Chat with a standardized message suggesting antibiotic de-escalation if clinically appropriate. Our primary endpoint was antibiotic duration and time from procalcitonin result to antibiotic de-escalation. Differences in mean times were compared using paired t-testing. 

Results: 

We conducted a nudge intervention on 48 patients between August-December 2023, and identified 48 patients between June-July 2023 to serve as a baseline comparison group. The mean antibiotic duration was 105.84 hours and 123.24 hours for the pre-intervention and intervention groups respectively (p=0.29). The time from procalcitonin result to antibiotic de-escalation was 105.85 hours and 108.15 hours for the pre-intervention and intervention groups respectively (p=0.87). In the intervention group, 25% of providers de-escalated antibiotics within 24 hours after nudge intervention. 

Conclusions: 

Nudge interventions did not influence clinician decision making regarding procalcitonin-guided antibiotic de-escalation, when evaluating mean antibiotic duration or time from procalcitonin result to antibiotic de-escalation. Despite high utilization of the procalcitonin assay at our institution, results rarely impact clinical decision-making. The utility of procalcitonin as a tool for antibiotic stewardship, absent more effective decision support, is questionable. 

Recommendations/Next Steps: 

Lack of improvement following a nudge implies that knowledge gaps do not explain discordance between testing and antibiotic prescribing. Future interventions could focus on guidance at the point-of-care through indication-based ordering protocols or other forms of restricted usage. 

Background:

Blood product transfusions are one of the most common therapeutic procedures performed in hospitalized patients. Currently, transfusion information is primarily available in UC San Diego’s electronic medical record, EPIC, in a “Transfusion” summary tab. Data available on this tab is incomplete and challenging to navigate. Furthermore, it is difficult to find pertinent blood product data and identify the status of blood product orders in an efficient manner, as data is dispersed throughout multiple locations in EPIC. As a result, delays in blood product transfusions are common. For instance, nurses and providers often call the Blood Bank directly to determine if ordered blood products have been prepared.  Not only does this increase the volume of tasks for clinical team members, the phone calls also disrupt the work of Blood Bank staff, affecting the turnaround time for all orders.  Consequently, blood transfusions may be delayed, at times with life-threatening or fatal consequences.  Another gap in the current EPIC layout is the absence of a quick way to determine if a patient has a history of red cell antibodies, transfusion reactions, or pertinent blood product restrictions.  Without this knowledge, providers may not be aware of patient-specific requirements or extended time required to procure compatible products, which may also delay patient care.   

Description:

The goal of this project is to create a “Blood Product and Transfusion Dashboard” within EPIC, to provide a more efficient way of gathering blood product and transfusion related data and, ultimately, to decrease delays in the transfusion workflow. This will be accomplished by synthesizing data populated in various areas of EPIC and presenting it to providers and nurses in a succinct and readily digestible manner (Figure 1). The dashboard will include pertinent hematologic lab data, with the patient’s hemoglobin, platelets, coagulation factors, haptoglobin, fibrinogen, and type and screen status. It will also include information regarding anticoagulant medications, presence of historical red cell antibodies, and if a consent for transfusion is on file. Beyond this, it will include more intuitive descriptions of product transfusion status: “Ordered”, “Being Prepared”, “Ready for Pickup”, “Transfusing”, or “Transfusion Complete”. Furthermore, the dashboard will leverage data that is actively entered into the patient’s medical record, such as when the clinical team scans a unit, initiates transfusion, and stops the transfusion.

Expected Outcomes:

Multiple benefits are expected with implementation of this dashboard, including decreased time from blood products being ordered to initiation of transfusion, decreased delays in initiating blood product preparation, and improved workflow for nurses, providers, and the Blood Bank staff when determining the status of a blood product order. Overall, these goals should lead to an increase in patient safety and fewer adverse outcomes related to delays in blood product transfusion.

Recommendations:

Next steps involve implementation of the dashboard, while obtaining both timestamp data from EPIC and user survey data to evaluate for potential decreased transfusion delays and improved user experience.

Background:

Primary care providers (PCPs) address many issues, including cancer prevention and screening, within the limited time available during appointments. Across the San Diego VA Healthcare System, rates of individuals who are up to date with cancer screenings are below the national VA rates. With PCPs being the first line of defense against screening for cancer, it is essential to address this gap in healthcare. This quality improvement project aimed to increase the percentage of up-to-date cancer screenings across multiple cancers (breast, colorectal, cervical, prostate, and lung) among patients at the VA La Jolla Primary Care clinic.

Methods:

We standardized workflow by incorporating cancer screening into the primary care note template to prompt the primary care provider to screen for breast cancer, colorectal cancer, prostate cancer, cervical cancer, and lung cancer. A customized after visit summary (AVS) that included cancer screening was created as part of this project and was given to the patient. The primary outcome was the proportion of cancer screenings completed between October 2023 and March 2024. We analyzed the trend of up-to-date cancer screenings across our patient panel from December 2023 to March 2024.

Results:

Through October 2023 to March 2024, 28 patients were found to be eligible and due for age-appropriate cancer screening, and 43 cancer screening tests were ordered. As of March 2024, 27 of these 43 (63%) tests were completed. Among the study sample, 5/5 (100%) of mammograms were completed, 5/9 (56%) of fecal immunochemical tests were completed, 1/8 (13%) of colonoscopies were completed, 14/16 (88%) of PSA labs were completed, 1/1 (100%) of Pap smears were completed, and 1/4 (25%) of low-dose CTs were completed. From December 2023 to March 2024, screening uptake improved in the three cancer types recorded on the Almanac database. Colorectal cancer screening increased from 48.6% to 50.2%, breast cancer screening increased from 70.4% to 76%, and cervical cancer screening increased from 69.8% to 72.6%.

Conclusions:

Incorporation of cancer screening into the AVS creates a tool that is patient-centered, and this tool has been distributed across the primary care clinic. However, further efforts are needed to increase completion of screening tests. The addition of an automated reminder for lung cancer screening is in progress, and future projects could study the effects of this reminder on screening rates. Completion of cancer screening requires steps outside of the primary care visit, such as scheduling with other departments, and may need to involve further interventions, such as sending a reminder to the patient if a test is not completed within a specific period of time.

Background: Medical errors are a major cause of morbidity and mortality in the healthcare industry. Human factors make errors much more likely to occur. However, explicit instruction on mitigating human factors is underrepresented in graduate medical education. The Naval Aviation community has systematically evaluated the risk posed by human factors and implemented specific approaches for reducing error, an intervention which has exponentially decreased preventable aviation mishaps over decades. Applying those principles from Naval Aviation to the way in which healthcare is delivered will reduce errors and improve safety.

Description of the Project: Human factors are a broad category within error analysis. Naval Aviation has grouped human factors into four broad categories: active factors, preconditions, supervisory factors, and organizational factors. Naval Aviation applies frameworks such as Threat and Error Management (TEM) and Crew Resource Management (CRM) to reduce the risks posed by these factors and prevent human error from contributing to mishaps. TEM and CRM directly apply to the medical industry and could immediately reduce dangerous mistakes that adversely affect patient outcomes.

TEM is a system designed to correct deviations from a standard. Deviations consist of threats, errors, and undesired states. Threats are risks which are known ahead of time (such as the risk of infection). We control threats by preparing for them appropriately (by giving perioperative antibiotics). Errors are harmless, but unintended, deviations that cannot be prevented (such as marking a wrong site for surgery). We control errors by identifying and repairing them (confirming the surgical site with the patient and perioperative team prior to induction of anesthesia). Undesired states are situations in which serious harm is imminent (such as excessive uncontrolled bleeding). We control undesired states by recovering from them (massive transfusion, IR embolization, etc). Effective TEM application requires the seven critical skills of CRM: Situational Awareness, Assertiveness, Decision Making, Communication, Leadership, Adaptability, and Mission Analysis.

We presented these principles to second year UCSD anesthesia residents as part of the professional development didactic series. We utilized analogies from real aviation disasters, description of aviation’s culture of safety, and group discussion on healthcare examples to illustrate how to apply TEM and CRM.

Lessons Learned/Expected Outcomes: 80% of respondents reported no prior knowledge of how the principles of CRM existed within their clinical duties. After presenting on TEM and CRM, 100% of respondents agreed that the information would positively influence their professional practice. Continued TEM and CRM training will improve team members’ ability to ensure patient safety in any situation or environment.

Recommendations/Next Steps: Educational emphasis on TEM/CRM, human factors, non-hierarchical communication, identification of cognitive bias, single provider CRM, and other Naval Aviation topics will empower members of the healthcare team and reduce preventable errors in patient care.

Figures/Tables: Diagram of TEM and CRM.