Anecdotal evidence from discussing with Gastroenterology providers at UCSF suggests that patient-provider messaging has become a major aspect of clinical care at UCSF, taking up significant time and providing valuable service to patients. However, there is no formal requirement to provide this type of care, nor is there a mechanism to provide re-imbursement. It is clear that this type of care will continue to increase in importance and use over time. I propose a project to formally study its use, highlighting the following aspects:

1) Assess its use and changes over time. Is the amount of messaging increasing year over year, and with an estimate of the amount of time spent on patient messaging on the provider level.

2) Quantify exactly what is being done in these messages. How many  are patient generated versus physician generated? How much messaging relates to new issues that were not brought up in clinic, versus follow-up of issues addressed in clinic? Is its use replacing clinic visits or augmenting clinic visits somehow?

3) Is there a away to assess the effect of its use? Can it cut down on patient visits? Are patients more satisfied with their care that use it? Is there a correlation between its use and physician burnout? Can we estimate a cost-benefit of its use somehow? Are there improved patient outcomes with its use? Can we understand better variation between patient populations and providers and its use? Can we extrapolate how it will be used in 5 years? Can we envision a mechanism for funding its use?

UCSF HIPSTER: Hypertension - Innovating Personalized STrategies for Excellent Results

In-person management of HTN poses a challenge for patients, physicians, the health system, and payers. It causes patients to incur the expense of missed work, parking and inconvenience and poses a barrier to care for a mostly asymptomatic condition.  For physicians, HTN management is often done on My Chart or by phone, as most of our providers do not have timely follow-up appointment access.  For our health system, visits for HTN reduce access for other types of patients who need evaluation and advanced testing.  Kaiser’s lauded HTN management service achieves excellent results, but employs expensive human capital (pharmacists, nurses repeatedly engaging with patients) almost exclusively to achieve their outcomes.  And for payers, office visits for HTN management is costly.  In terms of health care value, time to control and percent of patients controlled remains a challenge.  The new SPRINT trial, which will likely be incorporated into guidelines soon, will also raise the need for aggressive BP management with even greater precision, so as to avoid the complications of hypotension. 

To address these issues, it would make sense to leverage technology.  While apps and consumer medical devices can theoretically make personalized management more realizable, patient motivation to participate in self-care may be less for HTN than for more symptomatic conditions.  There is also a disconnect between smartphone app technology that can capture data and symptoms, the devices that record home BP, processes that would allow patients to communicate this information directly with their physician, and the workflow and incentives that would enable the physician’s office to act on this information.  This exposes a chasm and an opportunity in the current technology landscape to design holistic, patient-centered and physician-centered technology that makes the data actionable, personalized, and appropriate for current clinical workflows. 

We would like to design and pilot a portfolio of remote management programs that meet the needs of our various patients.  One example actively being designed by us will use an app and home blood pressure cuffs with an intelligent platform that engages patients ONLY when necessary, informs the care team ONLY when necessary, SIMPLIFIES communication using our own EMR, and enables SWIFT intervention by a care team. Rather than using technology extant, we are flipping the model and designing a product and workflow to fit UCSF health system and UCSF patient needs.  To do this, we have partnered with Vital Labs, Inc. and the UCSF School of Pharmacy to design a product for our needs.  Our goal is to continuously improve the technology based on our patients’ use patterns, user experience and feedback.  Technology often fails because it uses a heavy-handed approach, inundating patients with messages, texts, and tasks that are burdensome and not acted upon.  Our goal will be to continuously study the best possible prescription for BP measurement to result in the best outcome (BP at goal with minimal side effects using the most convenient medical regimen) without asking any more of our patients than we need. This rigorous evaluation can help us design an IT platform that overcomes, and doesn't create, healthcare disparities. Very preliminary assessment of an early version of the app in our clinic has shown that a diverse section of our population has been interested in adopting this type of management, including underserved populations.  Learning from this pilot, we will develop a HTN management menu to meet the myriad needs and abilities for our patients.  This may include traditional in-person visits, telephone e-visits, MyChart based communication, simple text messaging communication, and apps.  The aim is to devise HTN platforms that will enable us to prescribe technology judiciously, in just the right dose, individualized for each patient, and enabling physicians and clinics to operate with the utmost efficiency to achieve the best results.  We are collaborating with and learning from many divisions and units, including Primary Care, Nephrology, and Population Health, the CPI hub, and the CDHI, are very interested to hear from our colleagues' experiences at the VA and SFGH, and actively seek input from all corners.  We hope that our efforts will be generalizable to the broader UCSF patient community.  

Tech Challenge

Problems:  as highlighted via https://www.ucsf.edu/news/2015/06/130511/science-interrupted

1)      The buying power of the NIH dropped by ~$5 billion between 2003 and 2013 due to inflation.

2)      Paylines are 10%-14% for NIH dropping from 22% a decade ago

Solution:  Let’s crowdfund research.

Proposal:  Create a program to allow investigators to directly solicit crowdfunding support for their research projects.   Similar to Indiegogo and Kickstarter but for UCSF vetted scientific research.

This is a program that will supplement research support for investigators via a comprehensive crowdfunding platform.  This will involve a combination of mentorship and scientific advisory from senior researchers, crowdfunding from users and industry/venture capital, support from Department of Medicine (DOM) divisions, discounts with UCSF campus life for merchandise, marketing via UCSF branding, and other stakeholders.  A DOM crowdfunding platform will also provide a social networking component for investigators to provide updates on their research directly to funders as well as provide contact information for their labs via social media (twitter, facebook, instagram, linkedin, etc.).  These updates increase the potential for additional support after initial funding has been completed and increase visibility and transparency for labs.

Process:

1)      Investigators work with their chiefs/mentors to put together a brief one page proposal for posting to the DOM crowdfunding page.  They then upload their research proposal to the page for review.

2)      Submissions are reviewed by a DOM committee (e.g. senior investigators, industry experts, UCSF DOM leadership, DOM strategic management) for appropriateness and viability for crowdfunding.

3)      Accepted proposals then receive design and video support from DOM to create a unique crowdfunding page on the DOM platform.  Similar to https://www.kickstarter.com/help/handbook.  They also receive either matching division support or discounts to UCSF DOM merchandise for different tiers of pledged support (unique research t-shirt for $35 pledge, unique hat for $50 pledge, etc.)  This is subsidized by UCSF Campus Life who can assist with design and may provide close to at-cost pricing.

4)      If funding level is hit, the project will be funded and funds will be set up as unique gift accounts and used in support of the PI’s research (up to 10% for salary support).  A 5% gift account fee can be assessed.

Benefits:

1)      Benefits for an individual funder?

1. Funds will be raised via a non-profit entity and will be tax deductible.  Also unique merchandize like t-shirts, pins, etc., that are associated with the research labs.  (e.g. #SmithCancerLab)
2. Benefits of putting funds directly into research without exorbitant overhead costs as well as knowing that the dollars will go to a world-class researcher whose work and potential for success has been vetted by others at UCSF.

2)      Benefits for an industry/VC funder?

1. In addition to potential tax breaks, industry contributors get a chance to invest in the research of investigators who may one day take their research private.  Downstream, these industry/VC funders will have an early association with investigators who may wish to collaborate with the same industry partners who helped support their research careers.

3)      Benefits for PI?

1. Funding and exposure for research.
2. A platform for future funding as well as potential collaboration from other researchers
3. Potential for higher level collaboration with industry/VC as research progresses.
4. Creates a mechanism for PIs to solicit direct funding for their research from friends/family/others.

4)      Benefits for DOM/UCSF?

1. Increased funding for investigators
2. Visibility for specific research
3. Social networking for research
4. Leverages positives of the tech and social awareness climate and the current desire for innovation while ensuring research integrity

Draft budget:

$15k website and platform

$15-20k video production costs (for individual submissions)

$15k mobile

$10k marketing

$10k HIPAA compliant hosting and basic support for one year

The Problem. Ambulatory physicians utilize many different types of standardized assessments, ranging from depression screening, to functional pain assessments for patients on opiates, to incontinence and benign prostatic hypertrophy screening for older women and men. Many of these assessments are used in both primary care and specialty practice. Which assessments are done at a given visit depends on the specific patient’s demographics (e.g., age, sex), when their last screening was and if they are due for another, and sometimes the patient’s diagnoses or medications.

The Barriers. There are several barriers to conducting timely, standardized, assessments tailored to the patient’s needs. 1. Currently this requires an individual – often the physician – to investigate and note that the patient should have a particular assessment. 2) These assessments are often time consuming to conduct in practice. 3) Many assessments are available in paper instruments for patients to complete themselves, but one-in-four patients have inadequate health literacy, and assessments are not always available in the patient’s preferred language. 4) The results of the assessment require physicians to document them in the note or scan them into APeX, and then they are not easily found or tracked.

The Solution. A platform that allows information to be pulled from APeX (e.g., demographics, relevant medications or diagnoses), and sent in real-time to a secure device which can provide low-literacy, multi-lingual, video-doctor directed assessments to the patient while they are waiting to see their doctor. The tablet will then integrate with the patient’s visit encounter to send the assessment information into a flowsheet or the progress note so that the physician can see the results and act on them. Each assessment would also be linked to an educational video for those patients who identify a problem (e.g. for stress incontinence a video about kegel exercises, or for depression an introduction to evidence based treatments), which the patient can view before or after the visit depending on timing. These videos could also be made available on-line for patients with a secure password that would be generated by their answers to the assessment.

Such a secure, low-literacy, multi-lingual application already exists. It is called Phrazer and is being used in other settings already (emergency department, inpatient), but has not been used in ambulatory care. http://www.myphrazer.com/ The company that has developed Phrazer is interested in partnering to develop content for ambulatory use and clinical integration.

Needs. Although the application exists, there is still need for the following: 1) UCSF to develop a platform to pull information from APeX and push it to another application for tailoring of assessments – such an platform would be useful for other applications in the future; 2) selection of best practice standardized assessments that will be useful for the largest number and range of patients presenting to primary care, and that would also be useful in various specialty care; 3) content of video-doctor educational modules with translations and bilingual-bicultural video-doctors in UCSF’s main non-English languages; 4) integration to push assessment information into APeX; 5) workflow assessments for use in clinical practice; 6) pilot use in practice with assessment of patient, staff, physician acceptability and perception of usefulness, measurement of process improvement metrics. 

We propose the use of activity trackers (such as 'Fitbits') in an inpatient center to encourage patient mobility and more precisely measure the amount and types of activities patients are performing. De-identified data from these activity trackers would be presented electronically both to providers/staff and to patients in order to motivate patients and to allow staff to see patients that were due for a mobilization session. 

This project could use off-the-shelf technology with some minor modifications to enable it to be used in a health care setting and to enable data to be collected and displayed in a centralized location.

Our desired outcomes would be increased mobility of inpatients, stable level of functionality at discharge, decreased length of stay, decreased falls, and increased rate of discharge to home.

Background

As the population of hospitalized patients ages we will increasingly encounter patients with functional limitations of at least 1 ADL upon admission (1). During hospitalization these patients are vulnerable to the consequences of hospitalization and decreased mobility such as pressure ulcers, delirium, decrease in muscle mass, and ultimately an additional decline in their functional status (1).

In a study of patients aged 70 and older and with low or intermediate levels of mobility 29% had a decline in at least one ADL at discharge and 22% were discharged to a new institutional living situation (2).

Current Research

The STRIDE program at the Durham VA (3) has demonstrated that an activity program can lead to a decreased rate of discharge to skilled nursing facilities. A challenge that this program faced was monitoring and encouraging activity. There is some evidence that activity trackers can lead to a change in behavior and an increase in frequency and amount of activity.

Conclusion
Our innovation would be to bring this technology to the inpatient setting and display mobilization data in real time and in a centralized location. Receiving this grant would allow us to purchase equipment and enable staff to implement this project. This would then hopefully lead to improved functionality and health of our patients.

References

1.  Kleinpell RM, Fletcher K, Jennings BM. Reducing Functional Decline in Hospitalized Elderly. In: Hughes RG, editor. Patient Safety and Quality: An Evidence-Based Handbook for Nurses. Rockville (MD): Agency for Healthcare Research and Quality (US); 2008 Apr. Chapter 11. Available from: http://www.ncbi.nlm.nih.gov/books/NBK2629/

2.  Brown, C. J., Friedkin, R. J. and Inouye, S. K. (2004), Prevalence and Outcomes of Low Mobility in Hospitalized Older Patients. Journal of the American Geriatrics Society, 52: 1263–1270. doi: 10.1111/j.1532-5415.2004.52354.x

3.  Hastings, S. Nicole, Sloane, Richard, Morey, Miriam C.,Pavon, Juliessa M.,Hoenig, Helen, Assisted Early Mobility for Hospitalized Older Veterans: Preliminary Data from the STRIDE Program, J Am Geriatr Soc 62:2180–2184, 2014

There is increasing interest in the use of telehealth initiatives nationally.  Simultaneously, Medicare and other payors are increasingly recognizing that home-based care can be effective, cost-saving, and more person and family centered.  To date, opportunities for trainees to learn about home-based care are limited by current funding mechanisms.  For example, the primary care exception that allows residents to see patients in clinic and bill medicare, does not apply to the home setting. 

At UCSF, Primary care residents and fellows have had opportunities to see patients at home on a very limited basis via the UCSF Division of Geriatrics Care at Home Program.  However, this opportunity has been limited by lack of funding for faculty oversight, and most of the visits are done without capability for attendings to precept the visit LIVE.  The use of telehealth for trainee education could help provide funding for faculty oversight while also enriching the trainee experience by being able to have "live" precepting.  If funded, this opportunity could be avaiable starting with Internal Medicine Residents, Geriatrics Fellows, Nurse Practitioner students, and Palliative Care fellows.  If successful, it could also be easily expanded to include other disciplines.  For example, live telehealth consults for homebound patients who cannot otherwise see specialists, for example, orthopaedics.

This idea is innovative, interdisciplinary, feasible and has the potential to add a new dimension to our education mission, our trainees needs, and help provide person and family centered care. 

Physician note-writing is an opportunity to teach, share ideas, and solidify clinical decision-making.  Yet as billing requirements and copy-paste functionality have increased in recent years, the utility of writing and reading these notes has declined steeply.  In addition to the dehumanizing "tyranny of clicks and auto-populated fields," (Rosenbaum, "Transitional Chaos or Enduring Harm? The EHR and the Disruption of Medicine"; N Engl J Med 2015; Oct 2015; 373:1585-1588), we have all experienced the embarrassment of reading notes riddled with glaring, late night errors that become a permanent part of the medical record.  The bulkiness of auto-populated information in notes also reduces the reliance on written notes for inpatient communication.  While in-person verbal communication is critically important, the EHR is not meeting its potential as a reliable account of patient care, and the burden of the "tyranny of clicks" is detracting from our ability to produce notes that are as meaningful as possible.  Furthermore, the extent to which errors in notes contribute to poor patient outcomes at our center is not known.  In order to improve medical education, increase efficiency in medical care, and reduce avoidable errors, there is an urgent need to re-evaluate how we compose our inpatient progress and consult notes.

This proposal has three main goals.  First, we will track “typographical” errors in notes, comparing anonymized data by author type, service, time of day signed, and total number of notes written by that author per day. We hypothesize that errors will be prevalent, and that the number of errors will be higher among individuals who write many notes per day and sign their notes later in the day.

Second, we will characterize the teaching value of notes (graded by specific criteria) across services.  We hypothesize that the teaching value of notes will be inversely proportional to the bulkiness of excess information contained in auto-populated fields.

Finally, we will perform rigorous anonymous provider surveys to determine their trust in the EHR documentation.  We hypothesize that trust in the accuracy of EHR documentation is low, and that the majority of providers will view EHR documentation as a hindrance to rather than an aide for patient care.

These initial goals will lay the foundation for an intervention in which we will test two methods of note writing - free text versus autopopulated - and compare outcomes of documentation accuracy, trainee learning, provider satisfaction, and patient care.

Detailed Methods

Goal 1.  To track errors in notes.  A smart programmer (name TBD) will develop code to identify inconsistencies and errors in notes.  For example, when physical exams describe inconsistent findings that can never co-exist (for example, "obtunded" in the same exam as "neurologic exam nonfocal"); failure of a treatment or hospital day to be updated (same day "X out of X" of antibiotics listed for several days in a row); or list medications that have been discontinued (this will be complicated but should be able to be cross-referenced between A/P section and active med list on that day's note). The scope of errors in medical documentation will thus be defined as a starting point.

Goal 2. To characterize the teaching value of notes.  The burden of charting has taken away the joy of an important purpose of writing notes - reflecting on the patient and describing one's thinking process in words.  We will take a sample of notes from different author types (including consultants) and de-identify the author and patient information, then subject the notes to a grading system by which the teaching value of the notes is measured.  The criteria will include differential diagnoses, literature review, and descriptions of clinical reasoning.

Goal 3.  To assess provider trust in the EMR.  Using surveys targeted for different types of providers, we will assess the ways in which EMR notes are used – for verifying recommendations, understanding patients, and confirming plans.  We will solicit providers’ main sources of dissatisfaction with EMR documentation and their ideas for improvement – addressing “the third problem: many clinicians know what they want — but haven't been asked” (Rosenbaum 2015).

Achieving these goals has the immediate potential to improve clinical service and education.  Once the systems for tracking these errors and measuring teaching value have been developed, the project will be scalable beyond UCSF to other academic teaching hospitals that rely on EPIC.  The time frame for implementation will be feasible prior to December 31, 2016 as our goals are well-circumscribed and will be focused on inpatient documentation only.  Eventually, the intervention stage of the project will allow UCSF to be a pioneer in testing new methods of ensuring quality patient care through note-writing and transcending the drive for physician notes to be the sole source of billing documentation.

Important questions

What about outpatient notes?  This is a critical area with its own pitfalls and opportunities and thus will require its own plan for assessment and implementation, which can be addressed at a later time.  The scope of the current proposal is strictly for inpatient notes.

What if people start paying more attention to their notes because they know this is going on?  That would be great!  It would be a terrific outcome of the project and would not affect the overall purpose, which is to enhance the meaning and teaching value of the documentation included in the EMR.

What about billing???  This intervention will be the first step toward our combating the problem that a focus on billing has taken over our focus on patient care.  We have the opportunity to be at the forefront of necessary reform of the medical system causing "enduring harm," job dissatisfaction, and poor patient care.  This proposal will establish our leadership in implementing measures to ensure that physicians and our patients survive the “transitional chaos” successfully.

Proposal Overview:

Assessment of graduate medical education (GME) trainees is a high priority of training programs.  Training program directors must be able to certify that graduates are able to practice independently in their respective specialty/subspecialty.   Most Accreditation Council for  Graduate Medical Education (ACGME) programs, including the UCSF Department of Medicine (DOM) ACGME programs,  have an assessment system that is designed to report milestone assessments to the ACGME and the organization responsible for certifying individuals (such as the American Board of Internal Medicine).   Milestone assessments are assigned by programs for faculty or other supervising personnel to complete and return to the program for review by Clinical Competency Committees (CCCs).  The relevance of the milestones to the complexity of physicianship as well as trainee and faculty engagement with the assessment process has presented challenges to collecting, analyzing and reporting the clinical competence of trainees for DOM training programs.  We seek to create a novel, learner-driven GME assessment system, using Entrustable Professional Assessments (EPAs) in a newly developed information system that seeks to better assess trainee performance while empowering trainees to actively manage their own assessment process. 

Specific Aims:

1-     Design and implement an information system for all Department of Medicine educational programs that will collect and display trainee assessment data with customized views to allow trainees, advisors and competency committees to effectively assess trainee performance. 

2-     Change the culture of trainee assessment from program-driven to trainee-driven by enabling learners to pro-actively solicit faculty assessment in areas of need, monitor their progress with meeting assessment goals and to control the release of designated assessable elements to the clinical competency committee under the guidance of their faculty advisor. 

Background and Significance:

Various frameworks have evolved over the years to guide how to assess and report the competence of trainees.  The transition to a philosophy of assessing trainee outcomes began in 1999 when the ACGME introduced the six core competencies (patient care, medical knowledge, systems-based practice, practice-based learning, communication and professionalism).  The introduction of the Next Accreditation System (NAS) evolved these concepts to include milestones as subsections of the core competencies.  The internal medicine community began with 142 educational milestones which were subsequently reduced to 22 sub-competencies embedded in the original 6 ACGME competency domains.   (2)

In recognition of the limitations of the milestone-based assessment process an alternative approach to assessment uses EPAs as a framework for assessing competence.  EPAs reflect the degree of mastery of professional tasks that together constitute the work of a profession. (3)  Successfully performing EPAs requires abilities in multiple competencies.  Using EPAs for assessment recognizes that professional tasks are complex and provides a holistic view of the successful acquisition of professional skills. 

Current UCSF DOM Assessment Strategy:

In response to this rapidly changing landscape our current assessment strategy utilizes both milestone and EPA based systems to assess our trainees.  Our primary modality of assessment links specific milestones to trainee rotations and are sumatively assessed by faculty utilizing the E*value platform system.  Our programs may deploy different tools (global assessments, procedure observations, 360’s, chart stimulated recall, qualitative comments) to assess their program specific milestones but most programs depend upon observational methods to record the initial milestone assessment. Clinical competency committees (CCCs) then meet to determine the summative milestone score that is reported to the ACGME and the ABIM.

Currently, there are a few EPAs in use in certain programs (Internal Medicine, Geriatrics, and Hematology/Oncology) that are recorded in a Qualtrics survey platform but those data are hard to collate in a meaningful way for use by the CCC. Therefore, CCCs are largely dependent upon the milestone scores obtained from direct faculty observations, a process that has many limitations. .

Analysis of DOM Assessment Strategy:

Program CCCs are reporting issues with their milestone assessment processes that have been in effect since AY2013-14 for the internal medicine residency program and AY2014-15 for the fellowship programs.  First, compliance with completing the assessments is poor; across the DOM training programs, the overall compliance of assigned evaluations is initially only 50%.  Much administrative time is spent trying to get completed evaluations to the required 70% level and often program leadership has to step in to improve faculty compliance.  Reasons faculty cite for their challenges in completing assessments is “assessment fatigue” caused by multiple end of the month evaluations being due simultaneously, the need to assess as many as 22 milestones and overall frustration with the E*value platform.  Faculty also state that they have learned to ignore or delete automatic email assessment reminders from E*value, further complicating compliance with our evaluation process. Faculty also states that direct requests from trainees for feedback are harder to ignore due to the personal connection of the request. 

Second, the quality of the assessments is also in question.  CCCs are now often basing their judgments on insufficient numeric scores within each milestone category. This has led our CCC to become concerned about whether we are fully able to assess trainee’s performances.  Similarly, our CCC is unable to effectively identify areas of concern for struggling trainees, preventing appropriate remediation efforts. 

In an effort to promote EPA based assessment, the internal medicine residency began exploring the use of an EPA-based assessment system in 2013.  In 2013-4 we utilized a Delphi method to implement a group of 8 EPAs to help make our assessments more activity specific. (Table 1)  We also identified twenty-two subsequent EPAs with at least 80% participant agreement that can serve to guide our future expansion of this network. (6) 

Table 1.  Current EPAs housed on Qualtrics Platform

Although these and future EPAs contain of a robust menu of assessment parameters, we have been unable to implement them in a meaningful manner due to several limitations with our assessment technology as well as our inability to create a trainee-centered interactive assessment platform.  Table 2 outlines our efforts at implementation thus far. 

Table 2.  Timeline of EPA Implementation efforts

Vision for Future Assessment Process:

While learning about our aforementioned concerns regarding assessment amongst our faculty and CCC, we have been engaging our trainees to formulate our future assessment process.  Eight focus groups with a total of approximately 100 residents and fellows have been conducted over the last three years.  Several themes have emerged from these focus groups that match concerns voiced by resident trainees across the country when there is inadequate supervision or feedback.  (4-5)  These themes are summarized in Table 3.

Table 3.  Trainee themes from assessment focus groups.

To take an initial step of addressing these themes we began a two month pilot project where trainees actively solicit EPA activity-based feedback from faculty or other professional staff.  These evaluations are completely resident driven and are collected in Qualtrics. This “proof of concept” pilot enables us to examine the rates of completion of EPAs, willingness of residents to participate in the process and their satisfaction with the feedback they receive. This pilot is currently ongoing but has been well received by trainees and evaluators alike. 

Although this pilot is an exciting advance, it only fulfills our goals of facilitating timely trainee feedback. The pilot does not address our goals of creating an accessible trainee information platform where trainees are able to drive their own assessments and highlight their best work for the CCC.  It also only takes a small incremental step towards moving our assessment and feedback culture forward to be more open and meaningful.

Proposal Goals:

We plan to create an assessment system that addresses our CCC concerns of obtaining adequate assessment data for trainee promotion, trainee concerns of transparency, timeliness and quality of their assessments as well as faculty frustrations of “evaluation burn-out”.  Our proposal is to accomplish this by expanding EPAs as well as creating a user friendly trainee landing page to consolidate assessment data. 

Goal 1- To create a trainee “landing page” that enables residents to review their summative and activity-driven (EPA) assessment data with their advisors to facilitate understanding of their assessment data. 

Figure 1- Mock trainee landing page.

Goal 2- Give trainees more control over their assessment data that is presented to the CCC for entrustment.  Figure 2 demonstrates assessment modalities immediately available to the CCC.  Assessment data in red is available to the trainee and their advisor.  Through joint discussion they can release their best work for “entrustment” by the CCC.  For each EPA, trainees must release three “entrustable” evaluations done by three separate faculty members for evaluation by the CCC.

Figure 2- Types of data available to trainees, advisors and CCC

Goal #3- Improve timeliness, quality and specificity of assessment information for trainees while reducing evaluation burden for faculty by creating streamlined EPA assessment forms and handheld app to facilitate immediate completion.  The assessment editing will be driven by ongoing feedback from trainees and faculty from the current pilot project.

Goal #4- Utilize collected data from new information system to become more transparent about CCC deliberations with trainees.  With improved assessment data/information system the CCC will generate automatic entrustment emails for EPAs/procedures when trainees are signed off.  Similarly, we will be able to target earlier and more focused development efforts for residents who are not reaching their milestones appropriately.

Team Members:

Jeff Kohlwes MD, MPH.  School of Medicine.  Department of Internal Medicine.  Clinical Professor of Medicine. Director, PRIME Residency Program, Associate Program Director for Assessment and Feedback Internal Medicine Residency Program.

Patricia Cornett MD.  School of Medicine.  Department of Internal Medicine.  Clinical Professor of Medicine.  Associate Chair for Education, Department of Internal Medicine.

Vanessa Thompson MD.  School of Medicine.  Department of Internal Medicine.  Assistant Clinical Professor of Medicine. General Medicine Clinic Assistant Medical Director San Francisco General Hospital, Internal Medicine Residency Director of Academic Development

Sumant Ranji MD.  School of Medicine.  Department of Internal Medicine.  Associate Program Director for Quality and Safety.  Internal Medicine Residency Program.

Irina Kryzhanovskaya MD.  School of Medicine.  Department of Internal Medicine.  Senior Resident in Internal Medicine, Chief Resident-Elect.

Ray Chen MD.  School of Medicine.  Department of Internal Medicine.  Junior Resident in Internal Medicine

The Apex interface to find information on advance directives and POLST forms is very basic and deficient when searching for surrogate decision makers. In several instances the information is inaccurate or incomplete, leaving providers in the very difficult position of trying to implement the best interventions for each patient. This effort usually requires a substantial investment of time and resources in the attempt to clarify important directives with the patients or to search for decision makers. The aim of this proposal is to improve the accuracy of the current Apex interface and to create a new one to find surrogate decision makers. Such goals requires several steps: 1) Education of health care providers to conduct timely and adequate conversations with patients and families; 2) Streamlining the process of obtaining and entering advance directives and POLST forms in Apex; 3) Create in Apex an efficient search system for surrogate decision makers’ name and contact information. This will be the scholarly and QI project focus within the Social Work Fellowship in Palliative Care.  A social worker will be hired for advanced interdisciplinary training in Palliative Care and will assess and address the systemic issues related to documentation regarding surrogate decision makers and advance directives, enhancing the health care team’s ability to negotiate complex medical decision making with patients and families. He/she will closely work with different health care providers, including faculty, residents, fellows, nurses, unit clerks and IT analysts. 

Time is money and timesheets are an essential procedure to appropriately calculate the number of hours employees have worked. Accurate and updated timesheets are legal documents and essential for protection in the event of potential litigation  over unpaid wages or missed meal breaks. If there is an issue with
payroll, a timesheet will be needed to verify the employee’s hours.  Timesheets are your “evidence” in the event of litigation and provide  all parties with a written record that can be used to fairly resolve  wage disputes.Timesheets are necessary for tracking hours, and calculating overtime when it is due. The timesheet indicates the time the employee begins work; the time they go to lunch; the time they  return from lunch and the time they leave work.  In many cases, especially in the labs there is a failure to complete timesheets, which result is inefficiently run labs and % efforts towards work on a grant that cannot be monitored.  It would be good to have either a card reader in the lab to track the work hours or an app for a smart phone that will track the hours and record the GPS liocation.

Co-authors: Michelle Mourad and Kara Bischoff

Background:

Adverse drug events are important preventable causes of hospitalization in older adults, however compliance with a complex medication regimen after discharge home is known to be challenging for patients and particularly for the elderly. 

Objective: The aim of this proposal is to use technology to create easy to use color-coded charts with medication lists in large characters and with pills’ photographs. Technology would also be used to send the medication list to the patient’s PCP.

Methods: Interface APeX with a patient information medication information database to create a patient centered medication list. Potential information not currently included in discharge medication lists would be: Pill description, indication for use (antibiotic, blood thinner, control diabetes, etc.) as well as the creation of a tailored medication schedule of what to take in the morning, afternoon, evening and bedtime. 

Outcomes evaluated: Impact on patient care would be evaluated using the current post-discharge phone call program already in place. Patients are currently asked whether they have any questions about their discharge medications and if they have filled their new prescriptions. Secondary outcomes would be all-cause readmissions and readmissions for medication adverse events. 

Co Authors: Kara Bischoff & Sirisha Naranaya

Background:

Advanced care planning documentation is recognized as an important tool allowing patients to direct the type and intensity of medical care they receive. Systematic and educational innovations over the past decade have been effective at increasing the conversations around advanced care planning as well as the numbers of completed advanced care planning documents such as advanced directives (AD) and Physician Orders for Life Sustaining Treatments (POLSTs). 

Objective:

The goal of our proposal is to standardize advance care planning documentation and to create a centralized location in the chart for this documentation that is easily accessible across settings and to all clinicians caring for a patient. 

Description:

We reached out to a broad group of stakeholders across many disciplines and work areas including Primary Care, Oncology, Symptom Management, Home Based Primary Care, Housecalls, Population Health, Inpatient Palliative Care, and the Transitional Care and Heart Failure Program and identified the following needs:  

1. Standardized Documentation: Templates and/or dot phrases to guide documentation of best practices for advance care planning
2. A Central Location: One universally accessible location, such as a tab in Chart Review, where all inpatient and outpatient goals of care conversations and advance care planning documents can be quickly and easily found. Given the need for communication across transitions, a central location could also ensure that this documentation could be easily shared with partnering facilities across transitions in care.
3. Clear Display of Key Information: A prominent location in the current outpatient workflow tohighlight important information, such as patients’ surrogate decision-makers andprevious code status orders, in a visually prominent place in ApeX across all care settings (e.g. devotinga portion of the “Snapshot” to Advanced Care Planning)
4. Accessibility of Information Across Care Settings and Disciplines: Ensure that clinicians from all disciplines have the ability to document information about goals of care and advance care planning and have access to the same information.
5. Discharge Summary: Improve documentation of goals of care in hospital discharge summaries

Based on these needs we propose the following interventions:

1. Standardized Documentation: Improve templates and/or dot phrases to guide documentation of goals of care: 

• Create a dot phrase and/or note template for use in the Code Status and Advanced Directives note type
• Create a dot phrase that can be used in the overview of Problem List problems “Goals of Care Counseling and Discussion” and “Advanced Care Planning”
• Make these notes accessible to multiple disciplines, including social workers and nurse

2. Chart Review Tab: Create a central location where inpatient and outpatient goals of care conversations can be found

• Change the tab in Chart Review from “AD/POLST/Legal” to “ACP/AD/POLST” or to “GOC Documentation”
• Within this tab include:
• All notes with the “Patient Care Conference” or “Code Status and Advanced Directives” note type
• Problem List notes for the “GOC Counseling and Discussion” and “Advanced Care Planning” problems
• Consider inclusion of “Significant Event” notes
• Consider creating a new note type called “Goals of Care and Advance Care Planning”, which could be templated as above, to use for all such discussions
• File scanned ADs and POLST documents in this tab

3. Snapshot: Create a section of the “Snapshot,” commonly used in the outpatient setting, where the presence of ADs, POLSTs and prior inpatient Code Status can be quickly seen (this is present in the banner bar in the inpatient setting). Make sure this is present for all clinical disciplines in all settings.

4. Discharge Summary: Improve the “Expressed Wishes” section of the current inpatient discharge summar

• Make this documentation synergistic with documentation in the Problem List, ideally by pulling in the overview of the “GOC Counseling and Discussion or “Advanced Care Planning” Problem List notes.

5. Audit and Feedback: A report has already been created that allows us to audit inpatient documentation of advanced care planning. This will be used to feed data back to teams. 

It is a common perception on campus that "Epic is closed and doesn't want to talk to outside systems."

While there is a kernel of truth in this observation, Epic actually has several standard, robust Web services--and additional customized ones written by our APEX developers--through its interconnect API. After undergoing appropriate security review, YOUR application idea can connect to UCSF's electronic health record in a way that lets it read the patient roster, the treatment and care teams, and even vital signs and some lab values. This data can be used for everything from syncing lists to creating a "second skin" over Epic for specific clinical leads like a nephrology clinic or endoscopy suite.

In that spirit, we are proposing an APEX HACKATHON. Bring your developer friends, come prepared to write code, or just come with a great idea about what you could do with access to certain parts of the EHR IN REAL TIME. Want to see a beautiful graph of medication burden for your clinic panel? Want to track and plot your patients' immunosuppressant levels X days out from transplant, and build in logic for when a level is or isn't therapeutic? No idea is too hare-brained. Spend 24 hours writing code and building an application that connects to one of the practice, non-PHI environments in teams, getting practice for when/if you want to do the real thing.

At the end of the Hackathon, participants will submit their developed ideas for demonstration and the best idea will win a cash prize and the opportunity to develop the idea further in the UCSF environment. Any code developed at the Hackathon is owned by the developers.

The Hackathon will be great for advancing the digital health space for DOM, not least by building partnerships with developers and within our own institution, and for giving our incredibly smart community a taste of what can be achieved with what Epic already allows.

If the DOM funds this proposal as part of the IT challenge, we'd use the money to plan and put on the event, including giving out prizes.

Thanks for liking and commenting on our proposal!

Raman Khanna, MD

Aaron Neinstein, MD

Priyanka Agarwal, MD

The problem: Hospitalized patients get fewer than five hours of sleep a night.[1]

Poor sleep leads to increased rates of delirium, falls and hypertension, and to lower patient satisfaction scores and longer hospital stays. [2]^{, [3], [4]} Nighttime disruptions – including vital signs, lab draws and medication administration – are a major contributor to poor sleep. These clinical activities can often be shifted to waking hours without compromising patient care.

However, Apex is not currently able to show physicians the orders that will occur at night, or to give physicians the option to easily change them to waking hours.

That’s where Bed Rec comes in.

How it works: Prior to signing out for the day, a provider opens the Bed Rec tab. Under this tab, Apex will list the active orders that will take place during “sleeping hours” from 10pm to 6am.  Providers will then have the option, if clinically appropriate, to cancel these orders or shift them to waking hours.

Examples of Bed Rec in use:

Vital signs

• Bed Rec shows vitals are scheduled for Q4 hours, taking place at 8pm, 12am and 4am.
• This patient’s vitals have been stable and suspicion for clinical deterioration is low, so the provider can change vitals to 10pm and 6am

Medications (switch to PRN)

• Bed Rec shows Zofran is a q6 hour “standing” order and is scheduled at 8pm, 2am, 8am
• The provider can change the 2am dose to “PRN” while keeping the daytime doses “standing"

Medications (switch to TID)

• Bed Rec shows amoxicillin is written for Q8 hours and the patient will be woken to receive a dose at 4am
• This medication can safely be given three times a day during waking hours rather than strictly every 8 hours, so the provider can change it to TID and schedule a 6am dose.

Labs

• Bed Rec shows that the patient is scheduled for a 4am BMP and CBC.
• The provider can change the labs to 6am.

Obstacles and Concerns:

1. Patient Safety

If we monitor patients less closely at night, will this compromise our ability to recognize clinical deterioration? To address this concern, the Bed Rec tool will have validated decision support built in. The Modified Early Warning Score (MEWS) uses recent vital signs to identify patients who are at high risk for an adverse event or clinical deterioration.[5] Patients with MEWS scores of one or less have adverse event rates of just 5.0 per 1000 patient-days, compared to patients with MEWS of seven or greater, who have adverse event rates of 157.3 per 1000 patient days.[6]

On the page where nighttime vitals are listed, the MEWS score will also be listed in green, yellow or red, and labeled low-, medium- or high-risk for clinical deterioration. Providers will be advised not to forego vitals for the ~50% of patients with a MEWS of greater than one.

2. Will busy residents and attendings really take the time to use this tool?

One of the most common complaints we hear when rounding on patients is that they slept terribly. If the interface is user-friendly and intelligently built to allow for changes with the fewest number of clicks, residents will view this as a “high-yield” use of time. They will be able to give patients eight hours of uninterrupted sleep (a treasured commodity in the medical field) in a minute or less.

Lose the paper surveys. They waste everyone’s time from the receptionest to the patient to the records clerk to the physician. The Questionairre could be accessed from the home computer through the 'UCSF 'my chart application' and it would become instantly paperless and quick reading for the clinician when preparing for the patient visit.

This application could save time and instantly become part of the patient record.  This would save staff time in checking in patients, save patient time and would get rid of 1950 style paper records.

For those that do not use computers, offer kiosks with touch screens. 

Key Team Members:  Cheong S. Ang, MS, MBA, Grace Chen, MS, Lisa Hsiao, MArch, and George Su, MD

Problem Statement: Healthcare is complex, and some key intricacies come from linkages between stakeholders, agents, and dependencies. These are further complicated by care adjustments, iteration, inconstant human behavior, and inherent health and safety implications. Healthcare delivery can be distilled, however, as in other industry sectors, down to relatively simple processes. Despite a widespread strategic focus on process in healthcare, process management in healthcare remains profoundly underdeveloped. Dominant existing tools, such as the electronic medical record (EMR), are designed to house static data, and less to manage, map, or codify active processes. Thus, the healthcare system is relegated to managing healthcare processes with ad hoc workaround solutions, in addition to EMRs, such as spreadsheets, email prompts, calendars, and project management platforms—not to mention high-effort solutions like checklists and registries.

Innovation:  We’ve developed a novel, low-code application that organizes and contextualizes healthcare processes into a care pathway map, so that all users – from the healthcare team to the patient – are aware of and accountable to the healthcare plan. This involves the distillation of main healthcare processes into a kit-of-parts of “Actions and Nodes” which organizes workflows, tasks, manipulated objects (e.g. documents and records), roles, and agents into a timeline of tasks, dependencies, and results. This inherently flexible application transforms the volume of data for an individual into a coherent patient story. It also creates a powerful tool that can be used to optimize workflows, clinical care, and patient engagement.       

Solution description:

A.  Care pathway map

• A user-centered synthesis of process elements (e.g. “patient journey” or “administrator workflow”) maps Actions and Nodes into an IoT (“Internet of Things”) mesh network.
• The execution of Actions can be manually or automatically documented.
• Nodes are junction points that, depending upon clinical inputs (tests, labs, diagnostic studies, provider decisions), can trigger new actions in a patient’s overall care pathway.
• Users, including “agents”, “team members”, “stakeholders”, “patient” etc. are assigned “roles” and relevant preferences.
• Add-on functions, such as task listing, reminders, prompts, document management (faxing, e-signatures), and communication modalities (email, SMS), are configured to function within Actions and Nodes.
• Customized task lists are sortable by due date, priority, responsible party, care path destination, care path status, activity level, risk category, patient identifiers and characteristics.

 B. Analytics package for Process Management

• Identifies key metrics to measure achievement of goals, outcomes, and efficiencies
• Constructs “intelligent care pathways” through business intelligence and process goal targets
• Compares different pathways, clinics, and systems through benchmarking
• Provides a simple way to generate special reports, e.g. Chronic Care Management (CCM) effort reporting

Potential Impact:  A legitimate process management tool for healthcare has promise to dramatically improve provision of patient care. Efficiencies are realized by automating manual tasks and eliminating redundancies. Accountability and safety are enhanced by enforcing priority levels, driving task completion, maintaining pathway momentum, closing follow-up gaps, and raising attention to vulnerabilities. Patient involvement is encouraged and patient satisfaction is increased through new engagement opportunities at Actions and Nodes. Healthcare systems will have the agility to adapt to changing conditions, leverage newly-transparent processes to improve operational decision-making, and utilize new tools to intelligently manage process-related costs.   

Current solution status:  An IoT mesh network platform prototype and a minimally viable product (MVP) have been built (see attachment for sample screenshots). We seek funds to continue software development and to support evaluation of a pilot project based at San Francisco General Hospital. We also seek UCSF product development expertise and counsel.

Authors: Catherine Lau, James Harrison, Sarah Imershein

Background:

UCSF Health is committed to being the best provider of health care services and has dedicated many resources to improving the patient and family experience. Although the focus of improvement efforts has been communication coaching for providers and improving the provider experience, UCSF Health and other healthcare organizations have also begun leveraging information technology (e.g. MyChart, OpenNotes) to further empower patients and families with the goal of improving overall patient experience and health outcomes.  

The idea of further leveraging information technology to improve the patient and family experience also comes from patients and families themselves. A recent patient and family focus group strongly recommended that UCSF build a smart phone application that would allow them to be better informed about their care while being hospitalized. A separate Patient and Family Advisory Council at UCLA also made this recommendation and took this a step further by working with their local IT experts in developing an easy-to-use, intuitive, and informative smart phone application that is currently available for download from the iTunes store. The UCLA application has received rave reviews by patients, families, and healthcare providers.

Proposal:

We propose that the Department of Medicine develop and build a UCSF Health smart phone application with the aim of enhancing existing provider communication with patients and families while also better supporting patients and families when they receive care at UCSF Health. The initial app development will include adult inpatient medicine and any ambulatory practices or other departments/divisions that wish to participate. As the needs and wishes of inpatient and outpatient providers and patients will be different, separate “landing pages” for each clinic or division would be built within the app.

 Different sections of the smartphone app will include:

• Healthcare team members: This would include the multi-disciplinary healthcare team’s pictures and short biographies. The information listed here could be similar to that of existing paper provider facecards.
• Common conditions / diseases: This would include information on common conditions /diseases that are treated by that particular clinic or inpatient medical team as well as the usual, expected treatment plan. This section will require vetting from clinical liaisons to ensure accurate information and will also need to be written at an appropriate patient literacy level.
• My Treatment Timeline: This section would outline pending lab or imaging tests that their provider(s) have ordered as well as the timeline for completion of the tests. Active medications would also be listed here.
• Contact information for clinic or nursing station
• Nearby ATMs, restaurants, pharmacies
• Directory of Hospital / Clinic Services: This would include brief descriptions and contact information for spiritual care, café/cafeteria hours, gift shop hours, interpreter services, and service excellence.
• Patient Notes: This would be an area where patients can record healthcare provider voice instructions or take pictures using technology already on all smart phones.
• MyChart: A direct link to MyChart to allow users to access their medical record.

Feasibility:

As mentioned above, a very similar smartphone app has already been developed and built with the direct input of patients and families at UCLA and can be downloaded at this link: https://itunes.apple.com/us/app/ucla-neurosurgery/id1056278075?mt=8. We have relationships with individuals at UCLA who were involved in building and launching the app, and can help ensure that the UCSF build team does not have to start from scratch in this ambitious project.

Linking APeX to the app to allow for “My Treatment Timeline” and “MyChart” sections above may be a harder challenge, but has been done at UCLA with success.

Lastly, there will need to be a publicity campaign to ensure that patients and families are aware of the app.

Evaluation:

We propose implementing this technology as a randomized control trial that would involve inpatient medicine teams and ambulatory clinics that wish to participate. Teams and clinics would be randomized over a 3-6 month period with half promoting the use of the UCSF Health smartphone app while the other half would continue to promote the use of MyChart alone (existing standard of care). We would collect qualitative outcomes of patient/family and provider survey data and quantitative outcomes using HCAHPS and CGCAHPs Likelihood to Recommend Scores.  If a RCT is not feasible to implement, then a pre-post test design will be used. 

Current Challenges

We used to be able to place patient samples (primarily blood and tissue specimens from hospitals/clinics) and research samples between campuses using the existing shuttle system. This presumably has been halted since shuttle drivers cannot be held responsible for the potentially biohazardous contents of any packages and the subsequent procedures for handling the samples should something happen during transit. The halting of this resource has placed the task of specimen/sample delivery on lab technicians, clinical research coordinators and other researchers. This amounts to at least one full hour of lost productivity for each delivery (to and from the site of origin), creating an enormous personnel burden for the university staff.

Mobile App Based Solution

If a mobile app were to link the UCSF database of training history with the GPS location of a participating shuttle rider, the occupied shuttle could now be tagged as active for sample transport. The app would require UCSF personnel to sign up, linking their training history to their username and enabling GPS features of their mobile device. App usability would be broken into two phases: a data acquisition phase and a rider input phase. The data acquisition phase would simply utilize participants shuttle activity and training history to build a database of statistics on shuttle routes. This data would give you a percentage chance of each shuttle route being active for transport, defined by having at least one rider that is certified for handling biological specimens. UCSF would have to provide some basic incentive for downloading the BioCourier App, or simply incorporate this first phase of data collection using the existing UCSF mobile app. The second phase would begin to involve rider input. This would allow a rider to let the app know that they were about to ride a shuttle. This would raise certainty of shuttle activation to 100% and allow other researchers to plan their shipments more effectively. The timing of a passenger notifying the app of a ride is essential to labs getting notification for dropping off the specimens, so an incentive from UCSF would have to be involved to motivate users to accurately report their rides ahead of time. An accuracy rating for each user would be reported by the app so that potential shippers could see the likelihood that a rider would actually be present on the shuttle. This phase would work in tandem with the first phase, so if someone were to fail to show for a shuttle ride they activated, it could still be activated by another rider and the odds of that happening would be reported to the shipper. Incentives could include anything from discounted coffee and food on campus to bonuses through UCSF resources such as gym membership. If the app is well designed from a UX perspective, participation level required to ensure functionality would be quite low and alleviate a substantial personnel strain on labs and clinical research teams. Initial discussions with numerous PIs suggest that this sort of system would be enthusiastically and widely utilized across the university.

Objective

To create patient education materials and handouts for patients in their native language.

Background

San Francisco has an incredibly diverse population. Census data shows that approximately 45% of households have a language other than English spoken at home.  Nearly 30% of these households identify as speaking English “not well” or “not at all” resulting in approximately 1 out of 8 San Francisco residents identifying as having limited English proficiency (LEP)

Brief description of intervention

Despite the prevalence of LEP patients in our city, the patient education materials that we distribute to patients are often given only in English. This includes:

a)     the patient instructions and education materials when patients are discharged from the hospital

b)     the after visit summary given after outpatient visits

c)     the entire MyChart interface

d)     letters mailed to patients

e)     advanced directive forms available for inpatients at UCSF

Given the availability of automatic translation tools (granted of varying quality), we should be doing much better at delivering patients language concordant information. This proposal aims to identify better ways to provide patients materials in their native written language. This effort could be implemented in many ways

a)     As an consideration in all new projects (including many of the proposed projects on this site) such as:

                        i.         the PREPARE project (providing multiple language guidance through advanced care planning)

                       ii.         the automatic dictation of visit summary project

b)     As an addition/revision of already existing resources including the ones listed above (patient education materials, after visit summary, mailings to patients)

In terms of implementation, there would be a number of potential approaches including:

a)     Automatic translation of English words typed by a provider (much like how Google does)

b)     A dictionary of common phrases that are already translated into the three to five most common languages. These can be used by providers trying to communicate common instructions / directions

c)     Translating many of our already existing patient education and informational materials (or finding similar ones already in the patient’s language) and making it accessible in Apex to providers trying to provide patient education whether in the discharge instructions, patient's after visit summary, or via MyChart