Www.umich.edu



Recommending Strategies for Implementing Virtual Care to Improve New Patient Access in the Taubman Center Surgery Ambulatory Care UnitUniversity of Michigan Health System 19F6 Final Report Submitted to:Angela Haley, Administrative Manager, TC Surgery Ambulatory Care UnitSarah Forton, CI Fellow, Michigan Medicine Continuous ImprovementJeff Steinke, CI Specialist, Michigan Medicine Continuous ImprovementAndrew Sweeney, CI Specialist IE, Michigan Medicine Continuous ImprovementMark Van Oyen, IOE 481 Professor, Industrial & Operations EngineeringMary Duck, IOE 481 UMH Liaison, Michigan Medicine Continuous ImprovementSubmitted by: IOE 481 Team 6Megan Rapuano, IOE 481 StudentGarrett Schaub, IOE 481 StudentChristian Shields, IOE 481 StudentLexie Waltz, IOE 481 Student Date Submitted: December 10th, 2019TABLE OF CONTENTS TOC \h \u \z EXECUTIVE SUMMARY6INTRODUCTION PAGEREF _jq14084qchku \h 6BACKGROUND AND KEY ISSUES PAGEREF _g7lrg9vynl1p \h 6Clinic Overview PAGEREF _5nkh0fl93jkc \h 6Current State Problem PAGEREF _k4u4h9s93wbq \h 7Impact of Surgical vs. Non-Surgical Candidates PAGEREF _9buctx9c3r98 \h 8Impact of Appointment Style PAGEREF _t0vykrhfeh7o \h 9Key Issues PAGEREF _j7rhccbfplxd \h 10GOALS, OBJECTIVES, AND EXPECTED IMPACT PAGEREF _664pb32pguie \h 10Goals and Objectives PAGEREF _co1kxes1e9hm \h 10Expected Impact PAGEREF _1pb8jjvgl2ff \h 11PROJECT SCOPE PAGEREF _fl9c08h0xu17 \h 11DESIGN PROCESS PAGEREF _xeovjl7m11b3 \h 11Engineering Challenges PAGEREF _5mmkwr2yx8j1 \h 12Literature Search PAGEREF _o56a1sja2k87 \h 12Deliverables and Design Tasks PAGEREF _hxy992csz2wy \h 12Design Requirements PAGEREF _361k0bvx6jw1 \h 12Design Constraints PAGEREF _7sbh92ojvoxv \h 13Design Standards PAGEREF _4421bh85pzu9 \h 13DATA COLLECTION AND ANALYSIS METHODS PAGEREF _sd27g5381zfa \h 13Historical Data PAGEREF _c6a7jnlh5e4l \h 13Patient Surveys PAGEREF _xyif8ecrru1k \h 14Provider Preferences PAGEREF _qrf4282n29ez \h 15Literature Search PAGEREF _6uheq45b1ues \h 15FINDINGS AND CONCLUSIONS PAGEREF _2mq60npquxa6 \h 16Historical Data PAGEREF _tyglxieqsxjx \h 1675% of patients exceed the two-week access time PAGEREF _9er37t2cr5hi \h 16Access time varies by diagnosis PAGEREF _iqb5nraiemja \h 16Over 50% of the major diagnosis category patients are not going to surgery PAGEREF _2b1vwfxa2ay2 \h 17Patient age does not seem to be a significant indicator of surgical candidacy PAGEREF _injmyd5ju0h \h 18Post-op video visits take less than half the time of in-clinic post-op appointments PAGEREF _glhfwplvvqn1 \h 19Patient Surveys PAGEREF _6nztkccbwgyw \h 20Distance travelled by patient impacts patient virtual care preferences PAGEREF _ktyf9jdiz7v8 \h 21Involvement of a primary care provider (PCP) in patient care may impact virtual care preferences PAGEREF _ar6aekar13xd \h 22Patients’ initial access time and desire to be seen sooner may not impact virtual care preferences PAGEREF _hy26tiyekbo1 \h 22Provider Preferences PAGEREF _kdybd5a58sxx \h 23Scheduling restrictions may impact access time PAGEREF _wiuonbuir0 \h 23Splitting provider preferences by columns reduces cognitive load and errors PAGEREF _jqsycp5ij55a \h 24Provider coverage imbalance impacts access time PAGEREF _6qsvb2syqbhp \h 24Literature Search PAGEREF _jcv2xvgtdq66 \h 25Virtual care has been successfully implemented in post-operative settings PAGEREF _dd2eovmqdlti \h 25Patients treated as a production level can reduce access time PAGEREF _kg6hru9myze1 \h 25Limitations of findings and conclusions PAGEREF _rcrd0lbr0fvf \h 25ALTERNATIVES CONSIDERED PAGEREF _5lkps03psnn6 \h 26Criteria for evaluation PAGEREF _yrdqitv840us \h 27Alignment with Michigan Medicine's virtual care task force's practices PAGEREF _ww1tua4hwcro \h 27Improve patient care standards PAGEREF _ppqgn3w85pox \h 27Alignment of implementation timeline with division goals PAGEREF _8yi4r8xp67kb \h 27Increasing two-week patient access PAGEREF _fewo4cqn7c26 \h 27Increasing overall patient access PAGEREF _nzri7gpdpu8p \h 27Provider acceptance PAGEREF _3mohjxek0dei \h 27Patient acceptance PAGEREF _cpq2k5yh4ygx \h 28Decision Matrix PAGEREF _8ghd5322domj \h 28RECOMMENDATIONS PAGEREF _fglb5t11z8yw \h 28FUTURE WORK PAGEREF _1q31khvrk3td \h 31REFERENCES PAGEREF _t3d67qffhp4w \h 33APPENDICES PAGEREF _bsgjnrl3k52m \h 34Appendix A: Requirements, Constraints, and Standards Matrix PAGEREF _61dr7wmrysqu \h 34Appendix B: Historical State Data Analysis Diagnosis Categories PAGEREF _cukm672d03yh \h 36Appendix D: Patient Survey (Front and Back) PAGEREF _htoafqir251r \h 38Appendix F: Patient Survey Analysis Summarized Findings PAGEREF _wjhare6msic1 \h 43Appendix G: Provider Preferences PAGEREF _ubwgfgxnh6na \h 48Appendix H: Pugh Selection Matrix PAGEREF _bdyhuiaa2jji \h 48 TOC \h \z \c "Table F -" LIST OF FIGURES AND TABLES TOC \h \z \c "Figure" Figure 1. Patient Care Timeline PAGEREF _Toc26884990 \h 8Figure 2. Comparison of Average Access Time by Provider and Diagnosis PAGEREF _Toc26884991 \h 17Figure 3. Patient Access Preferences PAGEREF _Toc26884992 \h 20Figure 4. Survey Questions Regarding Patient Virtual Care Preferences PAGEREF _Toc26884993 \h 21Figure 5. Virtual Care Opinions Compared to the Distance Traveled for Appointment PAGEREF _Toc26884994 \h 22Figure 6. Virtual Care Opinions Compared to the Involvement of a PCP PAGEREF _Toc26884995 \h 22 TOC \h \z \c "Figure D -" Figure D - 1. Patient Survey Front Page PAGEREF _Toc26884996 \h 38Figure D - 2. Patient Survey Back Page PAGEREF _Toc26884997 \h 39 TOC \h \z \c "Figure F - " Figure F - 1. Patients' Preference to be Seen PAGEREF _Toc26884998 \h 43Figure F - 2. Patients' Distance Traveled Compared to Access Preferences PAGEREF _Toc26884999 \h 44Figure F - 3. Patients' Desire to be Seen Sooner Compared to Access Preferences PAGEREF _Toc26885000 \h 45Figure F - 4. Patients' Estimated Access Time Compared to Access Preferences PAGEREF _Toc26885001 \h 46Figure F - 5. Patients' PCP Involvement Compared to Access Preferences PAGEREF _Toc26885002 \h 47 TOC \h \z \c "Figure G -" Figure G - 1. Example of Provider Diagnosis Preferences PAGEREF _Toc26885003 \h 48Figure G - 2. Example of Provider Preference Column Updates PAGEREF _Toc26885004 \h 48 TOC \h \z \c "Figure H -" Figure H - 1. Pugh Selection Matrix PAGEREF _Toc26885005 \h 49 TOC \h \z \c "Table" Table 1. Average Access Time for New Patient Appointment for Significant Diagnoses PAGEREF _Toc26885006 \h 16Table 2. Percentage of Consultations Resulting in Surgery for Significant Diagnoses PAGEREF _Toc26885007 \h 17Table 3. Percentage of Patients that go to Surgery by Age Group PAGEREF _Toc26885008 \h 19Table 4. Percentage of Diagnoses Seen by Each Provider PAGEREF _Toc26885009 \h 25Table 5. Percentage of Patients to be Converted to Virtual Care PAGEREF _Toc26885010 \h 29Table 6. Percentage of Consultations Resulting in Surgery for Significant Diagnoses PAGEREF _Toc26885011 \h 29 TOC \h \z \c "Table A -" Table A - 1. Design Requirements, Constraints, and Standards PAGEREF _Toc26885012 \h 34 TOC \h \z \c "Table E -" Table E - 1. Percentage of Patients with a Two-Week Access PAGEREF _Toc26885013 \h 40Table E - 2. Average Access Time of Patients in Days PAGEREF _Toc26885014 \h 40Table E - 3. Percentage of New Patient Appointments that Result in Surgery PAGEREF _Toc26885015 \h 41Table E - 4. Average Wait Time for Surgery After New Patient Appointment in Days PAGEREF _Toc26885016 \h 41Table E - 5. Average Time to Wait for Post-Op Appointment in Days PAGEREF _Toc26885017 \h 41Table E - 6. Time of In-Clinic vs. Video Post-Op Appointments PAGEREF _Toc26885018 \h 42Table E - 7. Percentage of Patients that go to Surgery by Age Group PAGEREF _Toc26885019 \h 42 TOC \h \z \c "Table F -" Table F - 1. Count of Patients' Preference to be Seen PAGEREF _Toc26885020 \h 43Table F - 2. Patients' Distance Traveled Compared to Access Preferences PAGEREF _Toc26885021 \h 44Table F - 3. Patients' Desire to be Seen Sooner Compared to Access Preferences PAGEREF _Toc26885022 \h 45Table F - 4. Patients' Estimated Access Time Compared to Access Preferences PAGEREF _Toc26885023 \h 46Table F - 5. Patients' PCP Involvement Compared to Access Preferences PAGEREF _Toc26885024 \h 47 EXECUTIVE SUMMARYIntroduction and BackgroundIn the Taubman Center (TC) Ambulatory Care Unit (ACU), seeing patients for a consultation appointment within two weeks of booking an appointment is a high priority, but has historically not been achieved. The period between when a patient books an appointment and when the patient is seen for an initial consultation is referred to as “Access Time”. With a hospital-wide goal of providing 80% of patients with a two-week access time, the TC ACU lags far behind with only 25% of patients meeting the two-week access benchmark. The ACU was looking to mitigate the long access time and increase the number of patients that could be seen for a consultation within two weeks. As such, the administrative manager of the ACU engaged a student team from IOE 481 at the University of Michigan to assist in improving patient access time. Many divisions exist within the ACU, but according to the administrative manager, minimally invasive surgery (MIS) has historically been the division that consistently struggles to provide patients with a two-week access time. The administrative manager asked the student team to provide current state metrics on access time for the MIS division and to identify ways in which patient access time can be improved in the future. More specifically, with a clinic-wide focus on virtual care, the administrative manager specifically asked for recommendations on how to best implement virtual care to aid in decreasing patient access time. Patient access is affected by a number of factors, including, but not limited to, surgical candidacy, appointment style, and clinic scheduling practices. All patients, whether or not they will move onto surgery, are seen in the clinic for consultations. Meeting with all patients, regardless of surgical candidacy, contributes to a high demand for appointment slots, and thus increases the time that all patients must wait to be seen in the clinic. The ACU has historically seen patients in a traditional in-clinic setting, but the administration will be employing “virtual care” appointments to relieve some of the demand on in-clinic practices. These virtual care appointments include engaging primary care physicians to monitor care, and meeting with patients via video conferences in lieu of an in-clinic appointment. Lastly, providers’ preferences for the types of patients they will see limits the availability of some providers, and thus increases the time that patients must wait to meet with a provider.The administrative manager of the ACU engaged the student team to help the division improve patient access metrics, and as such, the goal of this project was to develop strategies for virtual care implementation in order to ultimately improve patient access. More specifically, the implementation strategies were to help the ACU administrators identify which types of patients would be best to transition to virtual care. These strategies were developed with the goal of ultimately contributing to a decreased patient access time. In support of the goal, the objectives of this project were to provide current state metrics of new patient access and to summarize patients’ preferences for virtual care implementation. MethodsThe student team focused on designing strategies for implementing virtual care. To do so, the team designed a series of data collection and analysis tasks to aid in the creation of recommendations. Historical Data Analysis The main component of design and analysis the student team conducted was a data analysis of historical clinic and operating room data. The data was compiled from the MiChart database and covered a span of 18 months, from January 2018 to June 2019. The student team received de-identified historical data from clinic administrators and worked with the administrative manager to synthesize the various diagnoses into 15 major categories. Next, the student team cleaned the data, removing records that were out of scope, and appending patient characteristic data to create a single, robust dataset. The student team then analyzed a variety of metrics slicing by provider and major diagnosis category. The student team analyzed metrics including, but not limited to, average access time, the percentage of patients with a two-week access, the percentage of in-clinic appointments that result in surgery, the average time patients wait for surgery, and the age of patients that go to surgery. The goal of the analysis was to understand where the MIS division stands in terms of patient access (Does access time vary by provider? By diagnosis? By patient characteristics? etc.) to ultimately inform which patients should be transitioned to virtual care.Patient Survey Design and AnalysisThe student team also designed and distributed surveys to MIS patients in the clinic over the course of three weeks. The purpose of the survey was to better understand if patients were comfortable and willing to participate in virtual care. As the surveys were voluntary, the student team received 69 total responses over the three weeks. Pivot tables were used to identify relationships, or lack of relationships, in the data collected. The goal of creating, distributing, and analyzing the surveys was to use the findings about patients’ virtual care preferences to inform the development of virtual care implementation strategies.Provider Preferences AnalysisThe student team then used a table of provider preferences, which is currently used by the call center to schedule appointments, to better understand provider capacity. The provider preference table lists every diagnosis that the clinic handles, each provider in the clinic, and preferences specific to each provider regarding whether or not (and to what extent) that provider will see patients for that specific diagnosis. The student team filtered the table so only in-scope diagnoses were considered and grouped the diagnoses into the 15 major diagnoses categories. Next, the student team created a visualization to highlight the providers that see patients for each individual diagnosis. The goal of analyzing the provider preference table was to understand how provider preferences might impact patient access time, and ultimately use the findings to inform the final recommendations.Literature SearchLastly, the student team conducted a literature search. The literature search included an analysis of one article on how production leveling or “heijunka” can be used to improve patient access. The student team also investigated a second article that explored how virtual care for post-operative visits decreases the average time patients need to be seen and potentially improves the quality of patient care. The goal of the literature search was to find outside research supporting administrator’s beliefs on virtual care. The research could be used to inform final recommendations about how effective virtual care is in improving patient access times. Findings & Conclusions The student team summarized the findings from each of the design tasks below.Data Analysis FindingsFor the historical data analysis, only the four diagnosis categories with the largest sample sizes were evaluated when developing recommendations. The four major diagnosis categories were gallbladder-related, groin & testicular pain/bulge, hernia, and general pain/ bulge, and made up 92% of all MIS patients analyzed in the historical dataset. The student team found that 75% of patients do not have a two-week access time, access time can vary between 22-36 days, all of the major diagnosis categories had less than 50% of patients going to surgery, and patient age is not a significant determinant of surgical candidacy. The last significant finding was that post-op video visits took approximately half the time of a traditional in-clinic visit for patients. The limitations of the historical data analysis were that the lack of the students’ medical knowledge made it difficult to accurately assess the impact of the assumptions on the results. Another significant limitation was that the dataset was not robust in listing specific information regarding patient characteristics. This restricted the student team to only draw connections between patient age and outcome in the clinic. Patient Survey FindingsAfter collecting and analyzing the patient surveys, the student team found 48% of patients surveyed were comfortable with virtual care, and only 26% of patients were concerned about discussing personal health information over video conferences. A slight correlation was found between virtual care comfort level and distance traveled to appointments. After 75 miles of distance traveled there was a general positive trend with virtual care comfort level. In this same distance interval, there was a decrease in concerns over virtual care as distance increased. Only 33% of respondents felt that they wanted to be seen for an initial consultation sooner. Provider Preferences FindingsAfter analyzing the provider preferences table, the student team found that the range of number of diagnoses a provider sees varied from 16 to 87 diagnoses. The team also found that provider preferences may impact patient access time, with certain conditions such as having an ostomy or scheduling restrictions impacting the most patients. These two things impact patient access time because the limited provider availability means patients with certain diagnoses will have to wait longer to be seen in the clinic. In addition, reorganizing the provider preference table to split provider notes into several columns can help call center agents by reducing the cognitive load, or the amount of mental work required, to sort through the table. Design AlternativesThe team designed three recommendations to aid in the implementation of virtual care and assist the MIS division in improving patient access time. The key criteria for the proposed alternatives in the project included alignment with standard virtual care practices at Michigan Medicine, improvement in patient care standards, alignment of implementation timeline with division goals, improvement of overall and two-week patient access, consideration of provider acceptance, and consideration of patient acceptance. A Pugh matrix was used to evaluate each alternative with the key criteria. After comparing each alternative individually, the student team found that a combination of the three would best accomplish the goals. RecommendationsThe student team was able to formulate several recommendations based on findings drawn from the data analysis, patient surveys, and provider preference analysis. The student team recommends converting varying percentages of new patient appointments to video visits or e-consults. For gallbladder related, groin and testicular pain, hernia, and general pain/bulge related diagnoses, the student team recommends increasing the percentage of patients seen via virtual care to 48%, 86%, 50%, and 82% respectively over the long term (approximately three years). These long-term goal percentages were found by using the current state rate of patients going to surgery from a consultation. Assuming that the rate at which patients with certain diagnoses going to surgery does not change, the student team developed this recommendation with the specific intention of transitioning non-surgical candidates to virtual care. In the shorter term, the student team recommends increasing virtual care percentage to 24%, 43%, 25%, and 41% over the next 12 months and 12%, 21%, 12%, and 20% over the next six months. The long-term percentages were calculated by adding a 5% buffer to the percentage of patients who currently go into surgery and subtracting that value from 100. The 12-month percentages were found by simply halving the long-term numbers, and the 6 months goals were half of the 12 month goals. Achieving each of these goals in an incremental fashion will allow the MIS division to ultimately decrease overall patient access time because fewer patients will be coming to the clinic.Second, the student team recommends conducting 50% of post-op visits via virtual care. The 50% target was based on survey results that showed almost half of all patients are already comfortable with the idea of virtual care. The student team also worked with the client to determine a realistic percentage of patients that could be transitioned to virtual care. For patients, virtual post-op appointments take significantly less time than in-person visits and converting would-be in-clinic appointments to video visits would reduce scheduling congestion and allow more open appointments for new patients. This in turn would reduce the overall access time for new patients.Finally, the student team recommends that provider preferences are managed in the provider preferences table. The student team recommends the clinic administrators work with providers to minimize the limitations on patient populations that each provider sees. By having providers see diagnoses they are not currently seeing, patient access time would be increased. This is something that is feasible given all MIS providers are trained for each diagnosis. Incorporating providers’ input into the process of reassessing diagnosis coverage is crucial to ensure this recommendation is smoothly implemented.INTRODUCTION The Taubman Center (TC) Ambulatory Care Unit (ACU) in the Michigan Medicine Department of Surgery is a clinic that provides General Surgery, Thoracic Surgery and Ostomy outpatient care for patients. An area of focus for the clinic is “new patient access,” hereafter simply referred to as “access.” Access is defined as the time that elapses between when a patient calls to schedule an appointment and when the patient is seen in the clinic for an initial consultation. After the initial consultation roughly half of patients schedule a surgery. The division of Minimally Invasive Surgery (MIS) within the ACU has a goal to see 80% of new patients within two weeks; however, the current percentage is roughly 25%, per the administration’s estimates. The ACU has historically seen patients in a traditional in-clinic setting, but has recently approved the use of “virtual care” appointments. Virtual care appointments will allow the ACU providers to video conference with new patients for a consultation in lieu of an in-clinic appointment. Virtual visits have only been tested with a select few patients, but the clinic intends to increase the frequency of these virtual visits in the near future. Because virtual visits are best for patients who are not immediately eligible for surgery, the clinic administrator asked the IOE 481 student team to specifically provide recommendations on strategies for identifying patients that should be seen via virtual care.Because there was such a large disparity between the current two-week patient access and the target, the focus of this project was to perform a current state analysis of new patient access metrics and identify ways in which the division can decrease the access time and increase the percentage of patients with a two-week access. Only the non-bariatric MIS division was the focus of the project, as the clinic administrators identified this division as having more room for improvement than Thoracic Surgery, Ostomy outpatient care, or other general surgery divisions.BACKGROUND AND KEY ISSUESThe following section will detail the background on the MIS division of the ACU and the current state problems that the division experiences.Clinic OverviewThe MIS division in the ACU clinic staffs seven physicians, three advanced practice practitioners (APPs), three registered dieticians (RDs), and three nurses. These providers see 24,000 patients every year for surgery-related appointments including, but not limited to, initial consultations, minor surgical procedures, and post-operative (post-op) visits. The clinic sees patients for a variety of minimally invasive surgeries including hernias, gallbladder removal, reflux surgery, bariatric surgery, lap band removal, and bariatric revisions. The entire division aims for a two-week access period, beginning from the moment the patient is scheduled for a consultation until they are seen for their appointment. Current State ProblemPatient access has been an area of focus for the division since the current administrative manager of the Surgery ACU assumed her role in April 2015. The clinic has hosted multiple continuous improvement efforts to address the patient access issues. The administrative manager notes that while overall access time has decreased, the two-week access percentage has remained relatively stagnant despite multiple improvement efforts. An important distinction should be made between access percentage and access time. Access time simply refers to the actual number of days that it takes for a patient to be seen for a consultation. Two-week access percentage (TWAP) refers to the percentage of patients that have an access time of less than or equal to two weeks. Both values are important to the division, and as such, they aim for a low access time, but a high TWAP. The values will hereafter be jointly referred to as “access metrics”. The MIS division specifically has been identified as struggling with their access metrics and has been chosen as the focus of this project. The current state timeline for a patient seen in the MIS division of the ACU is shown in Figure 1 below. In “Stage 1”, a patient is scheduled for an appointment. After that, the patient enters the “Wait 1” phase while they wait for the upcoming consultation appointment. This “Wait 1” phase is synonymous with access time. In “Stage 2”, the patient is seen by a provider and either deemed a surgical candidate or a non-surgical candidate (NSC). Once labeled a NSC, the patient leaves the MIS division and seeks care elsewhere. If deemed a surgical candidate, the patient faces another waiting period, “Wait 2”, before they have surgery in “Stage 3”. The candidate receives pre-operative care at a third-party clinic that does not impact MIS operations. Finally, the patient is seen in “Stage 4” for a post-op appointment. It is important to note that the student team did not attempt to mitigate the “Wait 2” period, as the administrative manager will be working on this as a separate project in the near future.Figure SEQ Figure \* ARABIC 1. Patient Care TimelineImpact of Surgical vs. Non-Surgical CandidatesOne way to improve patient access metrics is to be selective about which patients are seen in the clinic for an initial consultation. As stated above, patients are identified as a Non-Surgical Candidate (NSC) when they are seen for their consultation appointment. A patient may be deemed a NSC for a variety of reasons including poor health, outstanding health issues, or simply a diagnosis that is not handled by the MIS division. More specifically, the administrative manager originally noted that a patient’s BMI, age, and status as a smoker are the three key indicators that identify a patient as a NSC. Advanced analytics methodologies for identifying NSCs, such as machine learning, are out of scope due to the data limitations.When providers spend time in appointments with NSCs, it takes away time that the providers could have otherwise seen a surgical candidate. This is particularly impactful to the Surgery ACU because surgical specialties make most of their revenue from performing surgeries. The more NSCs a provider sees, the fewer surgeries they will perform, and thus less revenue for the clinic. Therefore, it is in the division’s best interest to minimize the number of NSCs seen in the office and use the majority of providers’ time to see surgical candidates. The administrative manager estimated that a large portion of patients seen in clinic are NSCs, but does not know the exact amount. The division has tried to identify NSCs at the time of scheduling and assign the APPs to these patients, but a large opportunity exists to divert more NSCs out of the clinic through other means. Impact of Appointment StyleFor consultation appointments, the MIS division hosts patients at the physical clinic in a traditional format where a physician or APP meets with the patient during the appointment. However, in an effort to streamline patient visits, the clinic has approved two different appointment styles that will be hereafter referred to as “Virtual Care”. The first appointment style is known as a video visit. Approved for both consultation and post-op appointments, video visits allow the patient to meet with a provider via a video conference hosted through the patient portal. The division has only trialed video visits with a select few patients, but is open to using these on as many patients as possible. The other appointment style is called an e-consult, which is a style of consultation appointment where the primary care physician (PCP) manages the pre-surgery care for the patient. When a patient is referred to the Surgery ACU, the referring provider would be asked to complete a questionnaire (created by the surgeon) with details to inform the MIS surgeons as to whether or not the patient is eligible for surgery. If the surgeon deems the patient a NSC, the PCP would work with the patient to improve their surgical candidacy or explore non-surgical treatment options. The MIS division would ultimately like to bring NSCs back to the clinic if they can become a surgical candidate. As such, it’s important for the division to use e-consults as a way to stay in contact with the patient’s PCP and not completely remove the patient from the division’s reach. If the PCP ultimately deems the patient a surgical candidate, the patient will move forward with a formal consultation, and will be given priority scheduling. While e-consults have not been used in the clinic before, they have been approved for use and mark a significant opportunity for the division. Managing NSCs’ care outside of the division would effectively free up MIS providers’ time to meet with patients who are more likely to follow through with a surgery.Patient SchedulingCall center agents schedule patient appointments via a provider table. The provider table contains a list of diagnoses along with the providers who currently see each diagnosis. Once The call center agent knows the diagnosis of the patient given by their PCP, the patient locates their diagnosis in the table and sees which providers see that specific diagnosis. Some patients may have characteristics which minimize the number of providers who are willing to see them, which limits the ability to schedule the patient and results in long access times. While there are medical justifications for many provider preferences, all providers are medically trained to see all patient populations. In the provider preference table, all information for scheduling preferences was stated under one column. From this, call center agents had to correctly identify whether the patient could be overbooked, which clinic was correct for the patient, if the patient had any pre-existing conditions that made them non candidates for surgery, and which provider could see each diagnosis based on the condition, provider name, and preference notes. Additionally, there was no standard order for how preferences were expressed. This made it extremely difficult for call center agents to accurately assess if a patient met the criteria to be scheduled for surgery. The provider table plays an integral role in how patient appointments are scheduled. Furthermore, current scheduling methods play an integral role in the bottleneck highlighted earlier. Effective changes made to the provider table may allow patients to be scheduled more efficiently, positively impacting patient access long term.Key IssuesSeveral key issues impact patient access time. The key issues of patient accessibility the team investigated are:75% of patients cannot be seen for a consultation within two weeks of requesting an appointmentNon-surgical candidates are a barrier to minimizing patient access times and maximizing revenue for the divisionCurrent scheduling practices lead to long patient access timesGOALS, OBJECTIVES, AND EXPECTED IMPACTIn working with the Administrative Manager of the Surgery ACU, a primary goal was identified and the primary objectives to achieve this goal were outlined as written below. The expected impact on key stakeholders is also an essential aspect of the project definition and is outlined below.Goals and ObjectivesThe outstanding goal of the division is to decrease patient access time. But because the student team had a short four months to implement changes that would result in improved access time, the student team’s goal with this project was to help the division develop strategies to achieve this goal. Specifically, the student team has sought to recommend implementation strategies exclusively for virtual care. These implementation strategies were developed with the ultimate goal of reducing patient access in the long term. The student team’s recommendation uses a historical data analysis and a patient preference survey to determine how to best implement new clinic practices to improve new patient access metricsThe objectives of the project were developed to support the goal of developing implementation strategies for virtual care. The two objectives below were used primarily to inform the final recommendation:Provide current state metrics for state of new patient accessSummarize patients’ preferences for virtual care implementationExpected ImpactThe team expects that their recommendations to make operational changes will reduce patient access time. Because the team is recommending strategies for implementation, the outcome will not directly be on the patient access time, but rather it will be on those who have the authority to make changes to clinic operations and scheduling practices in the near future. The recommendations for implementation will be targeted at the clinic administrators who will put the strategies into effect. The downstream expected impact from these recommendations is that patient access time will be reduced, and the overall patient experience will improve for the division. A potential negative impact of implementation of the recommendations is an increased backlog in OR access time for MIS patients. This impact has been discussed with the administrative manager and she is prepared to engage with a continuous improvement team to mitigate this impact.PROJECT SCOPE The main focus of the project was to understand patient access through data analysis. The following section details what was included and excluded from the scope. Scope includes:Access only for new and returning patientsMinimally invasive surgery (MIS), a subset of general surgeryClinic data only from the Surgery ACU Patients’ survey results regarding video visitsImpact of providers’ preferences on access Scope does not include:Bariatric patientsOther divisions in the Surgery ACU Operating room access metricsProviders’ feelings towards video visits and e-consultsIn-clinic operationsDESIGN PROCESSTo decrease patient access time, the team employed a large range of Industrial and Operations Engineering topics including, but not limited to logistics, data analysis, optimization, classification, and lean methodologies. Engineering ChallengesThe key engineering challenges of increasing patient access were as follows:Historical datasets were not be robust enough to analyze every metric with a fair sample sizeDue to narrow scope, some important data regarding patient appointments was filtered out of historical data analysis. This slightly skewed the findings, because the student team’s reported current state metrics do not holistically show the current stateLimited medical practice experience impacted the ability to develop recommendations that thoroughly considered medical nuancesBecause surveys were optional for patients, some questions had extraneous or blank responses (voluntary response biases), which impacted the ability to draw definitive conclusions about patient populations from the surveyStudent team had limited time to gather large set of quantitative and qualitative observationsDifficulties balancing optimized solution and quality patient careLiterature SearchThe team learned how production leveling and virtual care use can improve patient access through a literature search. The student team identified two articles that discuss how to improve access time by treating patients as a production level, and how virtual care can improve patient satisfaction in a clinical setting. Further details are discussed in the data collection and analysis methods section.Deliverables and Design TasksTo decrease patient access time, the following design tasks were completed:Conducted a literature search relevant to patient access and virtual careAnalyzed historical clinic and surgery patient data provided by the MIS divisionAnalyzed provider preferences to identify opportunities for virtual care implementation Created, distributed collected, and analyzed patient surveys regarding feelings towards virtual careThe design tasks informed the development of the final deliverable, which is a recommendation of implementation strategies for virtual care. The recommendation includes support from the historical data analysis, provider preferences analysis, and patient survey results.Design RequirementsThe student team considered the following key design requirements when developing the recommendation:The recommendation should increase two-week patient access and decrease overall patient access timePatients impacted by recommendations must feel comfortable with the care they are receivingProviders involved with implementation of recommendations should feel comfortable with implementationMore in depth information on the design requirements can be found in Appendix A.Design Constraints In designing the recommendations, the student team considered the following key design constraints:The student team had no medical knowledge so recommendations could only be data-drivenThe student team had to complete the project by December 10, so recommendations could not involve students past this datePatient care quality could not be compromised through the implementation of the student team’s recommendationMore in depth information on the design constraints can be found in Appendix A.Design Standards Throughout the project, the student team had to adhere to the following key design standards:Students adhered to HIPAA and University of Michigan medicine rules at all timesStudents used best practices when cleaning and analyzing dataThe team searched for more relevant standards beyond those stated above through an extensive online search but were unable to find more. On November 12th, the student team conducted a google search for key design standards related to this particular project, but were unable to find relevant standards. More information on design standards can be found in Appendix A.DATA COLLECTION AND ANALYSIS METHODSThe student team performed several data collection tasks throughout the course of the project. The following design tasks were completed to understand the current state of patient access and inform recommendations for improving the future state. Historical DataHistorical clinic and surgical data spanning January 2018 to June 2019 was de-identified and provided for student team. The student team used this data to identify current state metrics and identify patient populations that could be transitioned to virtual care. The team began by working with the administrative manager to group the 251 unique diagnoses observed in the data into 15 broader categories. A list of the major categories can be seen in Appendix B. Because the administration wanted to see the breakdown of patient access metrics by diagnoses, grouping the individual diagnoses helped ensure there was an ample sample size for each diagnosis grouping.After the diagnosis grouping was completed, the student team worked to clean the clinic data and match patients who were seen in the clinic to corresponding surgical records. To remain in alignment with the scope, the student team cleaned the data to only include Taubman Center appointments, non-bariatric patients, completed appointments, and other minor details to ensure accurate information would be reported. Because the clinic and surgical data were in two separate datasets, careful effort was taken to appropriately match up the surgical records to the correct clinic record by identifying patients, provider, and surgical dates that occurred after an appointment. Once surgical data was matched up to in-clinic appointments, the student team then took similar steps to appropriately match post-op appointments to patients who had surgery. Lastly, patient age and BMI from a separate dataset were merged into the complete appended dataset. A full list of data cleansing and analysis assumptions can be found in Appendix C. Next, pivot tables and charts were used to analyze the appended dataset. Using the pivot tables created from the appended data, the student team found current state metrics such as average access time in days, the percentage of patients with a two-week diagnosis, the percentage of in-clinic appointments that result in surgery, the average time patients wait for surgery, and the average time patients wait for a post-op appointment. Each metric was examined by provider and major diagnosis type. Patient Surveys The student team designed surveys to be distributed to MIS patients when they enter the clinic for both first time visits and return visits. The surveys were intended to gather information regarding patient scheduling preferences and if they did or did not want to be scheduled within the two-week standard scheduling goal. The survey also aimed to gain insight into patient comfort levels with virtual care, and if the time frame in which they would be able to be seen would impact their willingness to try this appointment style. The questions were created by the student team and reviewed with the administrative manager to ensure unbiased survey questions. The survey went through three iterations of review before the final survey, which can be found in Appendix D, was completed. The MIS division distributed surveys from October 28th to November 15th and received 69 responses. Surveys were given to patients upon check in and were collected when the patient was called back into the examination area. The student team reviewed and utilized pivot tables to better understand the relationships between the question responses. Conclusions were drawn from determining where clear relationships presented themselves, or lack thereof.Provider Preferences The student team analyzed provider preferences from the diagnosis table in order to better understand which providers had capacity to see new patients. First, the table was filtered so only MIS providers were being considered. Next, the student team analyzed each individual diagnosis and matched them to one of the 15 categories that was created during the historical data analysis phase. Next, the student team created a pivot table with individual diagnoses in rows and providers as the columns. The cells within the table indicated whether or not the provider sees patients with that diagnosis. Additionally, the student team took the preferences indicated in the “notes” section of the diagnosis table and created indicator columns of patient preferences that were either marked “yes” or “no” to align with currently used decisions trees for scheduling patients in the clinic. Literature SearchTo better understand patient access and the impact virtual care can have in a clinic, a thorough literature search was conducted, and two papers were found to be particularly relevant. The first paper reviewed, by Gavriloff [1], indicated that production leveling or “heijunka” in healthcare can help control timing of when patients are seen, which can improve patient access. This loading type can control the timing of patient visits and patient access, resulting in improved timely access to care in an ambulatory clinic. The main problem addressed in the article was how to make time for nonurgent care [1], such as new patient initial consultations. The team used this foundational work to show how new patient access can be modeled as a production level to be met by virtual care. A second study by Nikolian et al. [2] examined how virtual care can be used in post-operative care. The study found that there were no statistical differences in key metrics such as readmission and infection rates between traditional post-operative visits and video clinic visits [2]. The study also found that the use of virtual visits frees up more time for new patient visits, or four patients per week in this study [2]. The paper also revealed that there is a financial reason to implement virtual care, as it is less expensive for hospitals. Additionally, virtual visits are extremely beneficial to patients, saving time associated with travel to and from the hospital, eliminating in-clinic wait times, and saving money with travel-associated costs. The student team used this information to understand how virtual care could be implemented at the ACU, and which patients could potentially benefit most. FINDINGS AND CONCLUSIONSThe student team concluded the following key information from each data collection task and used these findings to inform the recommendations discussed later in this paper.Historical Data As discussed in the data analysis section, many metrics were explored using both provider and diagnosis category as filters. Due to small sample sizes of some major diagnosis categories, the student team decided to only analyze and consider the four diagnoses with the largest sample sizes when making recommendations. Gallbladder-related, Groin & Testicular Pain/Bulge, Hernia, and Pain/Bulge were the four major categories. These major diagnoses encompassed 92% of all patients analyzed in the dataset, so the major conclusions and forthcoming recommendations were based on these four major categories. The average patient access time for the top four diagnosis categories is shown in Table 1, but a complete listing of access time for all diagnosis categories and all other analysis results are listed in Appendix E. Table SEQ Table \* ARABIC 1. Average Access Time for New Patient Appointment for Significant DiagnosesIt is also important to note that because the scope of this project did not include bariatric patients, the sample sizes of patients seen per provider could be lower than the actual amount. Because of this, the student team elected to not use any of the provider-specific metrics in the development of recommendations. The major conclusions from the data analysis are summarized below.75% of patients exceed the two-week access timeAs the administrators suspected before the project started, a majority of patients are not being seen in two weeks, per the hospital-wide goal. As seen in Appendix E, in the historical state, only 25% of patients overall were seen within two-weeks of making an appointment. This was important to confirm, as being aware of the current state challenges allowed the student team to make realistic goals when providing recommendations for the division.Access time varies by diagnosisAs seen in Table E-2 in Appendix E, the average access time, across all providers, varied from 23- 36 days. As seen in Figure 2 below, for the four major diagnosis categories, the average access time varied by diagnosis for each provider. This was a significant finding because this meant that some patients waited, on average, a full two weeks more than others. The discrepancy was greater at the individual provider level. As seen below, patients seen by Provider 6 waited anywhere from 16 to 62 days-- for patients on the high end, that’s nearly seven weeks longer than the two-week access goal. Every provider sees patients for all of the four major diagnoses, which signifies room for improvement regarding distribution of patient types across providers in order to make access time even for all patients.Figure SEQ Figure \* ARABIC 2. Comparison of Average Access Time by Provider and DiagnosisOver 50% of the major diagnosis category patients are not going to surgeryAs seen in Appendix E, and highlighted in Table 2, the percentage of patients going to surgery after an initial consultation for Gallbladder-related, Groin Pain, Hernia, or Pain diagnoses were 47%, 9%, 45%, and 13% respectively. Table SEQ Table \* ARABIC 2. Percentage of Consultations Resulting in Surgery for Significant DiagnosesMajor Diagnosis CategoryPercentage of Consultations Resulting in SurgerySample Size (out of total in-clinic appointments)Gallbladder-Related47%49 / 104Groin & Testicular Pain/ Bulge9%10 / 108Hernia45%516 / 1154Pain/ Bulge13%11 / 84Out of all of the patients seen in the 18-month period analyzed, these patients represented 92% of all in-clinic appointments. As previously discussed, patients that do not go to surgery after an in-clinic consultation are not financially profitable for providers, and the use of consultation times on NSCs negatively impact the quality of care than can be given to surgical candidates being treated within MIS by decreasing access. Patient age does not seem to be a significant indicator of surgical candidacyAs previously noted, the division thought patients who smoke, are over 75, or have a very high BMI are not generally surgical candidates. By analyzing the percentage of patients that went to surgery out of the total patients seen in the clinic and segmenting the data by age group, the student team found that this didn’t necessarily hold for age (Note: the student team was unable to obtain data on patient smoking history, and data on patient BMI did not provide robust enough sample sizes for in-depth analysis). As seen in Table 3 below, the percentage of patients going to surgery did not dramatically drop off after age 75. Interestingly, the age group with the lowest percentage of patients resulting in surgery is actually 46-50 year olds. This information was imperative in informing recommendations regarding which patients could be transitioned to virtual care.Table SEQ Table \* ARABIC 3. Percentage of Patients that go to Surgery by Age GroupAgePercentage of Patients that go to SurgerySample Size<2067%2 / 320-2540%17 / 4326-3038%28 / 7331-3537%26 / 7136-4034%34 / 10141-4539%40 / 10346-5028%42/ 14851-5539%59 / 15356-6040%78 / 19461-6546%84 / 18366-7044%99 / 22771-7544%67 / 15476-8036%25 / 7081+35%18 / 51No data 6Post-op video visits take less than half the time of in-clinic post-op appointments When the student team analyzed the time that patients spend in the clinic compared to the time a patient spends when having a video visit, they found that the video visits take less than half the amount of time. The average time a patient spent in the clinic from check-in to check-out was 57.6 minutes across all diagnoses, but the average time a patient spent in a video visit was only 25.3 minutes. A full breakdown of time spent by diagnosis can be seen in Appendix E. It is important to note that the sample size of video visits analyzed was only eight because the clinic has only trialed video visits, compared to the 575 samples of in-clinic post-op appointments. However, the small sample size was still a good indicator of time that patients could save if they had a post-op appointment via video visit. While this data was not representative of time that the provider spends with a patient (because there are other things patients do in the clinic that don’t involve a practitioner), from a patient satisfaction point of view, this was a significant finding. Patient SurveysA primary reason for completing the patient surveys was to gain insight into whether or not patients were concerned about the inability for a majority of them to be seen within a 2-week access time. The survey revealed that only 33% of respondents felt strongly (indicating “yes” they would like to have been seen sooner) about decreasing their new patient access time. Of the respondents who felt strongly about decreasing their access time, 52% of them had access times greater than 1 month. It is also important to note that 51% of all patients either did not want to be seen sooner or had no preference regarding their access time.Figure SEQ Figure \* ARABIC 3. Patient Access PreferencesThe patient surveys also revealed key information regarding patients’ comfort levels with virtual care and how varying personal factors may influence those comfort levels. The findings revealed below were used to better understand what populations of patients may be candidates for the initial transition to virtual care. The full survey can be found in Appendix D, but the two questions regarding virtual care were phrased as seen in Figure 4. Figure SEQ Figure \* ARABIC 4. Survey Questions Regarding Patient Virtual Care PreferencesIn the analysis below, the label for question five was translated to “Virtual Care Comfort Level” and question six was translated as “Concerns over Virtual Care.” These two data points were averaged and then compared to every other question asked in the survey. Overall, the data showed that 48% of people had a virtual care comfort level of 4 or 5 (comfortable), and 74% of people rated their concerns over virtual care as a 1 or 2 (few concerns). This showed that about half of the surveyed responses were very comfortable with virtual care implementation but only approximately a quarter of people were concerned about discussing personal health information (PHI) over video conferences.Distance travelled by patient impacts patient virtual care preferencesAs seen in Figure 5, there was a slight correlation shown between the two variables, with patients who travelled under 75 miles responding with average answers suggesting they were ambivalent towards virtual care implementation. On the other hand, patients who travelled greater than 75 miles responded with average answers which suggested they lean towards supporting the implementation of virtual care practices. Patient’s virtual care comfort level had a linear R2 value of 55.99%, and patient’s virtual care concerns had a linear R2 value of 49.17%. For studies involving human behavior, R2 values of approximately 50% and above are considered adequate to infer correlation in linear models. Both of the values for this comparison therefore fall within an acceptable range.Figure SEQ Figure \* ARABIC 5. Virtual Care Opinions Compared to the Distance Traveled for AppointmentInvolvement of a primary care provider (PCP) in patient care may impact virtual care preferencesAs seen in Figure 6, the varying levels of PCP involvement seemed to show minor differences in virtual care preferences but due to small sample sizes this claim is limited. Interestingly, the patient population reporting “No PCP or N/A” were the only population to have their concerns regarding virtual care outweigh their reported comfort.Figure SEQ Figure \* ARABIC 6. Virtual Care Opinions Compared to the Involvement of a PCPPatients’ initial access time and desire to be seen sooner may not impact virtual care preferences As seen in Appendix F, the graphs showed minor trends among people who were seen within different time frames as well as people who had preferences about being seen sooner. Unfortunately, the sample sizes for the varying populations tended to be too small to definitely draw conclusions regarding whether a patient’s access time and access preferences would impact their virtual care preferences.Overall, the survey results revealed useful information to inform the student team on patients’ virtual care preferences but due to limitations of sample sizes, definitive conclusions regarding preferences of entire patient populations are limited and should be investigated further. The full data analysis can be found in Appendix F.Provider Preferences The following section highlights conclusions and findings from the analysis of the provider preference table regarding scheduling restrictions, cognitive load reduction, and the impact of provider imbalance on access time.Scheduling restrictions may impact access time Through our analysis, the student team found that provider preferences may impact patient access times. Specifically, the student team found several provider notes, conditions, and fundamental flaws with the table that made scheduling patients who are likely surgical candidates challenging. In addition, these recurring provider notes also impact access time since patients with preexisting conditions cannot be seen by some providers. Since these patients can’t be seen with these conditions, they would either have to wait for another provider’s availability or seek another method of treatment. Several recurring provider notes were identified for the MIS provider matrix: Patients with an Ostomy were often not allowed to be operated on Recurrent hernias were often not allowed to be seen without extra stepsOverbooking in the clinic was often regulated based on number of overbooks Additional conditions that made patients inoperable: cirrhosisheart failure being a transplant patientFurthermore, other aspects of the provider matrix that made scheduling difficult:Several conditions were able to be referred to other clinics instead Information on clinics was out of date Diagnoses were misspelled or had duplicatesThe student team found that these factors above posed challenges with patient scheduling. Patients who should not be seen in the clinic because they were not surgical candidates were often scheduled anyway. In addition, the student team found that a disorganized provider matrix leads to overall less efficient scheduling. Splitting provider preferences by columns reduces cognitive load and errorsTo remedy the issues of cluttered notes column section in the provider table, the student team created the following columns to reduce cognitive load to the call center agents and reduce clerical and scheduling errors: A binary (yes/no) overbooking allowance columnAn overbooking requirements columnAn ostomy allowance column A clinic preference column A provider preference columnBy specifying these columns, it was easier to see what the preference of the provider was versus only scheduling preferences. Decision trees with binary conditions are currently used in the scheduling call center, and incorporating these columns better matches the current scheduling process.Provider coverage imbalance impacts access timeNext, the team performed data analysis on the provider preference table. The objective was to find which provider saw each of the 228 diagnoses that were within our project scope (note: more diagnoses were listed in the provider table than the 215 observed in the historical data analysis). There were large imbalances in how many diagnoses each provider saw when categorizing individual diagnoses into the 15 major categories previously discussed. A visual data analysis was created for all diagnoses, to help the administration easily see which providers are willing to see patients with a specific diagnosis to help the administration easily see which providers are willing to see patients with a specific diagnosis. The full provider preference visualization can be found in Appendix G. Organizing the provider preference table in this way revealed that many providers do not see a number of the 103 unique diagnoses. The range of diagnoses seen by an individual provider varies from 16 to 87 diagnoses (excluding the APP). This highlights that there is a large discrepancy between the numbers of diagnoses that providers see. The student team concluded that this ultimately impacts access time because patients with diagnoses that only few providers see for have less of an opportunity to be scheduled in the clinic. Because all providers are medically trained to see all (MIS) diagnoses, this imbalance can be addressed. Table SEQ Table \* ARABIC 4. Percentage of Diagnoses Seen by Each ProviderProvider1234567 Number of diagnoses seen3234228763316 Percent of MIS diagnoses seen31%33%21%84%6%32%16%Literature Search Based on the literature search, two primary conclusions were drawn. Virtual care has been successfully implemented in post-operative settingsFrom the literature search on virtual care, it was apparent that virtual care is very successful in specific applications such as post-operative care. The student team used this literature to understand how virtual care can impact patient access and satisfaction. The recommendations for implementing virtual care were based on methodologies adapted in the paper. Specifically, the option to have video visits for post-operative appointments was found from this literature search. This article was also used as a reference to show that virtual care is not appropriate for all diagnoses, so it’s use will vary depending on diagnosis. Patients treated as a production level can reduce access timeTreating new patients as a production level would help reduce access time because appropriate resources would be devoted to new patients. Load leveling could be applied to virtual care practice by allocating virtual care appointments consistently throughout the day rather than in batches. By applying load leveling practices to virtual care, the clinic administrators will be able to further improve patient access than possible with virtual care alone. However, this should be considered as a later stage implementation technique.Limitations of findings and conclusionsThe limitations of the historical data were primarily due to the inability to access or connect some patient characteristics to the comprehensive analysis. Primarily, the student team was unable to link self-identified risks such as smoker status or a BMI of 40 or above to the data analysis and therefore was unable to draw definitive conclusions for why some patients/diagnosis groups have lower rates of being sent to surgery. Another key limitation addressed in the engineering challenges section was the nonexistent medical knowledge of the student team. There are likely reasons that certain practices or procedures for scheduling, assigning providers to patients, etc. may be carried out. With no medical knowledge of these practices or why they may be done, the student team was solely working off of insights drawn from the data analysis. As such, assumptions made (as listed in Appendix C) may not accurately represent the true nuances of the MIS division. The team worked closely with the administrative manager to address all assumptions, but it is possible that these assumptions may still have not accurately captured the intended metric in the dataset for all records. Lastly, because bariatric patients were removed from scope, the current state metrics are not representative of all patients seen by the MIS division.The primary limitation of the patient surveys was due to voluntary response biases and non-descriptive free text responses. Specifically, the question asking why the patient was being seen in the clinic resulted in minimal usable data due to lack of responses and inadequate descriptions of the primary diagnosis. Additionally, the sample size for some patient characteristics were too small to make definitive conclusions regarding differing populations. Lastly, there is a possibility that all respondents to the survey did not fully comprehend what the “video visits” fully entailed from the descriptions in the questions and therefore a limitation exists regarding the level of virtual care which respondents felt comfortable with. The surveys were able to provide a starting point to understand patient preferences regarding virtual care but a more in-depth study into comfort levels of varying patient populations would be needed in order to draw definitive conclusions regarding patient preferences in terms of diagnosis categories.A limitation of the provider preference matrix was that bariatrics patient data is a large part of MIS, and by removing it from the scope it was hard to tell which providers do not see many other diagnoses because they focus on providing patient care to Bariatrics. A limitation of the data analysis between the provider preference table and the historical clinic data was that diagnoses did not perfectly match. For example, the clinic data might have “left inguinal pain” listed, where the patient preference table might have “inguinal hernia pain”. This made it difficult to adequately categorize some diagnoses. ALTERNATIVES CONSIDERED The final recommendations to the administrative manager are intended to suggest strategies for how virtual care can be implemented in the MIS division of the Surgery ACU. The major recommendations focus on converting varying percentages of patients to virtual care, converting some post-op appointments to video visits, and managing provider preferences. Each recommendation was considered individually, but the student team decided a combination of the three best addressed the needs of the clinic. Other data collection tasks and design alternatives were not considered, but alternative variations to the recommendations were considered. The recommendations are based on the analysis of the historical data and the patient surveys, as well as the information gathered from a comprehensive literature search. The recommendation and alternatives were evaluated based on the criteria discussed below. Criteria for evaluationThere were several key criteria which the proposed alternatives were evaluated against, all of which relate back to the previously discussed design requirements and constraints. Alignment with Michigan Medicine's virtual care task force's practices An important factor for evaluation was aligning with Michigan Medicine’s virtual care task force’s practices because as the rest of the U of M health system transitions to utilizing virtual care, the MIS division’s plan must integrate into the health system’s plan.Improve patient care standards Improving patient care standards was important as well since the hospital as a whole is committed to serving their patients with the highest standards possible. While it was a hard constraint that the recommendation must maintain the current state of patient care, the recommendation aimed to improve the overall patient care experience which in turn could improve the patient care standards.Alignment of implementation timeline with division goals The alignment of implementation timeline with the overall division goals had been discussed with the administrative manager. The recommendation was evaluated based on the ability to help the MIS division reach their short- and long-term goals relating to decreasing patient access time.Increasing two-week patient access An increase in two-week patient access has continued to be the driving force and overarching goal in creating recommendations for virtual care implementation. This was therefore an essential criterion for evaluation of the recommendations.Increasing overall patient access An increase in overall patient access was also a driving force for implementing virtual care. Overall patient access was essential because it aligns with the division’s short-term goals and will lead to reaching two-week new patient access goals in the future. This was therefore an essential criterion for evaluation of the recommendations.Provider acceptance Acceptance by providers was also a consideration because without the support of providers, implementation was expected to be less successful. In order for a solution to be sustainable, the people implementing the changing practices must accept the changes and be committed to making these changes. Patient acceptance Acceptance by patients was the last consideration. While this criterion was similar to improving patient care, it was measured separately because it was from the perspective of the patient. Michigan Medicine could assure that implementation of a recommendation will improve patient care but due to personal preferences of patients there may be people who will be uncomfortable and unwilling to accept the change in patient care. If patients do not accept the change in their patient care the implementation will be unsustainable and therefore impractical as a recommendation.Decision MatrixThe recommendations that the student team created were evaluated based on the criteria discussed above in a Pugh Matrix shown in Appendix H. Each of the criteria were rated based on importance by a percentage, with all criteria adding to 100%. Each of the recommendations, including the current state as a baseline, were then given a grade from 1 to 5 for each of the criteria. The grades were multiplied by the importance percentages in order to get the overall scores for each of the recommendations. The first recommendation, discussed in further detail in the following sections, was to convert varying percentages of new patient appointments to virtual care, the second recommendation was to convert 50% of post-op appointments to video visits, and the last recommendation was to manage provider preferences. The Pugh matrix was used to compare the proposed recommendations in order to determine which recommendations would provide the most value for the division. Upon completion of the Pugh matrix, it was found that the recommendation to convert new patient appointments to virtual care would be the most impactful, followed by the recommendation to convert 50% of post-op appointments to video visits, followed by the recommendation to manage the provider preferences. The Pugh matrix revealed that all three recommendations would provide an improvement to the current state, and because all three can be completed in parallel to each other, the student team determined that all three alternatives should be provided as recommendations. These three recommendations will be discussed in depth in the following section.RECOMMENDATIONSThe following section outlines the recommendations the student team formulated based on information from the data analysis, patient surveys, and provider preference analysis. In addition, the following section outlines future work the student team believes could be performed. The following recommendations provide implementation strategies for virtual care and aim to ultimately increase patient access over time.Convert varying percentage of new patient appointments to e-consults and video visits Based on the information from the data analysis, the student team developed percentages of patients that should be converted to e-consults and video visits. As previously discussed, the student team found that over 50% of patients in the four most common diagnosis categories are not going to surgery. Based off of the data for the number of patients going to surgery in the current state, the student team developed the following percentages of patients that should be transitioned to virtual care in the next 6 months, 12 months and three years. Table 5 below shows the patients that should be converted to virtual care instead of being seen in the clinic for a consultation. Note that virtual care refers to both video visits and e-consults, but due to a lack of medical knowledge, the student team is not specifically recommending the breakdown of patients to be seen by each type of appointment. The conversion rate percentages were based off of the percentage of patients that historically went to surgery, repeated for ease of reference in Table 6 below.Table SEQ Table \* ARABIC 5. Percentage of Patients to be Converted to Virtual CareDiagnosis6 months12 months3 yearsGallbladder-Related12%24%48%Groin & Testicular Pain21%43%86%Hernia12%25%50%General Pain/ Bulge20%41%82%Table SEQ Table \* ARABIC 6. Percentage of Consultations Resulting in Surgery for Significant DiagnosesMajor Diagnosis CategoryPercentage of Consultations Resulting in SurgeryGallbladder-Related47%Groin & Testicular Pain/ Bulge9%Hernia45%Pain/ Bulge13%The percentages of patients to convert were calculated by adding a buffer of 5% to the current state percentage of patients who go to surgery. For example, only 45% of Hernia patients seen in the clinic today make it to surgery, so add a 5% buffer for error and the recommendation for the ultimate 3-year goal is 50% of patients should be transitioned to virtual care. The 12-month goal was found by dividing the 3-year goal in half, and the 6-month goal is slightly less than half of the 12-month goal. These measures were calculated by the student team to give the division a reasonable amount of time to implement the changes, with measurable goals to hit along the way.Using the findings from the patient surveys, transferring at least 50% of patients to virtual care is reasonable, as 48% of patients were very or extremely comfortable with having a video visit. The 26% of patients who showed concern over video visits, however, means that further education about the use, effectiveness, and safety of video visits are needed to ensure patient satisfaction does not wane. If the administration ensures that video visits are rolled out with the proper educational materials to ensure patient comfort, then the administration should be able to reach the varying goals in the specified amount of time. Patients who are likely not surgical candidates should be in the percentage of patients converted to virtual care in lieu of in-clinic consultations. The student team is recommending the percentage of patients to convert to virtual care, but is not restricting the method (e-consult or video visits) or which patients are candidates for virtual care. Many patients present unique conditions that providers judge on a case by case basis, and the student team is not medically qualified to influence these judgements. However, based on historic state findings, the student team does not recommend the division use age as an identifying characteristic of NSCs, as there was not a true trend between a patients’ age and likelihood of going to surgery. The Pugh matrix revealed that this recommendation would provide the largest impact on the clinic and should therefore be the top priority in implementation.Conduct 50% of post-op appointments via video visit As previously discussed, the student team found that post-op video visits take less than half as much time as an in-clinic post-op appointment. Converting half of patients to video visits for post-op appointments would result in greater patient satisfaction because patients would be spending less time overall in their post-op appointments. Additionally, patients would save even more time by no longer having to commute to and from the appointment. As seen in the patient surveys, 77% of patients are traveling more than 10 miles to the appointment. If patients only have to logon to the patient portal and speak with the provider from their own home or office, patients would not need to commute 10+ miles both ways, thus marking a significant savings in time and transportation costs. The student team determined that converting 50% of post-op appointments to video visit was a realistic metric, as almost half of the patients were already comfortable with video visits, per the patient survey data. That being said, further education would be required to ensure concerns about the safety and effectiveness of video visits are addressed for the considerable fraction of patients who still have some concerns.It is impossible to quantify impact to the division, because the only data available to the student team was time from check-in to check-out, and not specific time spent with providers. That being said, the student team reasons that converting half of the patients to post-op video visits would likely result in slight time savings for the clinic because the providers would simply need to meet with the patient for a video conference. Clinic receptionists, medical assistants, nurses, and others normally involved in the in-clinic care would not have to be involved in video visit care, thus resulting in a net time savings for the division as a whole.Manage provider preferences from scheduling capability The student team recommends that the clinic administrators work with providers to minimize the limitations on patient populations that each provider sees. As discussed in the findings section, there is a large discrepancy between the number of diagnoses that a provider sees, ranging from 16 to 87 diagnoses (excluding APPs). Having more providers see certain diagnoses such as cystic duct stones, carcinoid of the appendix, colectomy, gastric ulcer disease etc. can decrease access time because patients will have the opportunity to be seen by more providers. All providers in the MIS division are trained to see all diagnoses, this proposed change is therefore viable. Determining which diagnoses can be seen by other providers is beyond the student teams’ medical knowledge and the timeline of the project, therefore administrators should work with providers to determine the appropriate implementation of this recommendation. The analysis of provider preferences is limited by the unknown impact that bariatric patients have on providers bandwidth for seeing other diagnoses. In order to counter this limitation, the student team recommends that a yearly review is done on the provider preferences table to reduce errors from changes within the hospital systems.Additionally, to assist the call center agents in scheduling patients with the appropriate provider the student team recommends an alteration to the current provider table. Adding columns for overbooking and specific pre-existing conditions be created in the provider preferences table, as seen in Appendix G, could improve scheduling practices. By including new binary indicators for the notes call center agents would be able to more accurately schedule patients with physicians and it would improve the legibility of the provider table. Additionally, if the standardized columns were to be implemented across the hospital system, they would promote transparency in the scheduling system. FUTURE WORKThe work completed by the student team is intended to be a starting point for understanding the current state of the MIS division and present possible next steps towards the implementation of virtual care into regular practices. In the future, the division plans to engage with the Michigan Medicine Virtual Care task force in order to begin active implementation. This task force will be able to address the logistics of implementation and enforce the best practices which Michigan Medicine has outlined for virtual care.In addition, an expansion of the initial work done on the provider preference “matrix” to include bariatrics would be valuable to show access for all diagnoses in MIS. In turn, the full bandwidth of providers would be available, allowing gaps to be analyzed in provider care for the entire division. Columns could also be customized for bariatrics, with BMI being a crucial variable for correct scheduling. Additionally, as patient access time decreases, specifically for new patients, the Surgery ACU should consider engaging a continuous improvement team to address the possible increased backlog of surgical candidate patients whose OR access times could be impacted. In terms of patient satisfaction, it would be best if patients did not have to wait long periods of time to have their surgery after being seen for a consultation. Because patients come to the Surgery ACU to have surgery, the access time to the OR should be a primary concern for the division moving forward.REFERENCES [1]C. Gavriloff, S. Ostrowski-Delahanty and K. Oldfield, “Impact of Lean Six Sigma Methodology on Patient Scheduling,” Nursing Economics, July 2017. Available: ProQuest,. [Accessed October 25th, 2019][2] V. Nikolian, MD, et al. “Pilot Study to Evaluate the Safety, Feasibility, and Financial implications of a Postoperative Telemedicine Program,” LWW,Annals of Surgery, August 2018. Available: LWW, . [Accessed October 20th, 2019]APPENDICESAppendix A: Requirements, Constraints, and Standards MatrixTable A - SEQ Table_A_- \* ARABIC 1. Design Requirements, Constraints, and StandardsEntry #123RequirementsR-A. Organizational Policy(R-A-1)R-B. Ethical(R-B-1)R-C Health & SafetyN.A.R-D. Economic(R-D-1)R-E. Implementability(R-E-1)(R-E-2)R-F. User Acceptance(R-F-1)R-G. Patient Acceptance(R-G-1)R-H. Task Duration(R-H-1)Entry #123ConstraintsC-A. Organizational Policy(C-A-1)C-B. Ethical(C-B-1)(C-B-2)C-C. Health & Safety(C-C-1)C-D. EconomicN.A.C-E. ImplementabilityN.E.C-F. User AcceptanceN.A.C-G. Patient AcceptanceN.A.C-H. Task Duration(C-H-1)Entry #123StandardsS-1. HIPAA(S-1-1)S-2. Organization's Std.(S-2-1)S-3. Best Practice(S-3-1)S-4. Code(S-5-1)Standards that are not applicableNIOSHOSHARequirements:(R-A-1) The recommendation should have aligned with Michigan Medicine’s best practice standards for virtual care(R-B-1) The recommendation should have maintained current best practices for patient care(R-D-1) The recommendation should have aligned financially with virtual care implementation in like departments of Michigan Medicine(R-E-1) The implementation timeline given the divisions long term goals should have been considered(R-E-2) The recommendation should have increased the percentage of patients with a two-week and decrease overall new patient access time(R-F-1) The providers involved with implementation of the recommendation should be comfortable with the proposed next steps(R-G-1) The patients impacted by the recommendation should feel comfortable with the change in their medical care(R-H-1) The recommendation should have been designed with the most efficient methods possible to save time and laborConstraints:(C-A-1) Students had to complete all MLearning and compliance training prior to beginning data analysis(C-A-2) The student team only has access to a narrow scope of data (C-B-1) Patient care quality could not be compromised through the implementation of the student team’s recommendation(C-B-2) The student team had no medical knowledge so the recommendations could only be data-driven(C-C-1) Students had to obtain a flu shot by November 1 r to meet with client in the hospital (C-H-1) The student team had to complete the project by December 10, so recommendations could not involve students past this dateStandards:(S-1-1) HIPAA and Michigan Medicine compliance policies dictated the practices and standards students observed(S-2-1) The signed code of conduct dictated how students must acted and behaved in the hospital and in all professional interactions with client and coordinators(S-3-1) Students used best practices for cleaning and analyzing Excel data(S-4-1) Students received training in all fire code and emergency response codesAppendix B: Historical State Data Analysis Diagnosis CategoriesNoneHerniaGallbladder-relatedSpleen-relatedGroin & testicular pain/ bulgeEsophageal, non-herniaDiastasisPain/ bulgeAbscess/ infected wound/ fluid/ drainageVomitingObesityLipoma & subcutaneous noduleITPMassOtherAppendix C: Historical Data Analysis Major AssumptionsOnly completed new patient clinic data was analyzed, and all surgical and post-op data was only analyzed if a patient first had a new patient appointmentOnly new patient appointments at the Taubman Center in the MIS division, with non-bariatric patients were included in the datasetNew patient clinic data was sorted by newest first and matched to the oldest occurring surgical record for that patient/ provider combination; this meant the smallest amount of time in between a consultation and surgery would be reportedAccess time was calculated by subtracting the date the appointment was made from the date of the new patient appointmentDuplicate values were removed from surgical and postoperative data and examined manually to ensure that any duplicates were correctly matched to the appropriate new patient appointmentOnly “completed” or “scheduled” surgeries were counted when calculating the percentage of patients moving onto surgery or calculating the time that patients waited for their surgery after the consultation appointmentAny records that took over 1 year to book the surgery from the date of consultation were removed from calculations, as these major outliers were skewing the dataWhen calculating average time for in-clinic appointments and time spent on video visits, only TC locations were included and only appointments under 200 minutes were included, as the student team thought anything more than this was an unreasonably (and probably wrong) amount of time to spend in an appointmentPatient age was actually the age of the patient at time of analysis and not at the time of appointment (at most the age could be 1.5 years off from actual age at the time of appointment)Appendix D: Patient Survey (Front and Back)Figure D - SEQ Figure_D_- \* ARABIC 1. Patient Survey Front PageFigure D - SEQ Figure_D_- \* ARABIC 2. Patient Survey Back PageAppendix E: Historical Data Analysis Summarized FindingsTable E - SEQ Table_E_- \* ARABIC 1. Percentage of Patients with a Two-Week AccessTable E - SEQ Table_E_- \* ARABIC 2. Average Access Time of Patients in DaysTable E - SEQ Table_E_- \* ARABIC 3. Percentage of New Patient Appointments that Result in SurgeryTable E - SEQ Table_E_- \* ARABIC 4. Average Wait Time for Surgery After New Patient Appointment in DaysTable E - SEQ Table_E_- \* ARABIC 5. Average Time to Wait for Post-Op Appointment in DaysTable E - SEQ Table_E_- \* ARABIC 6. Time of In-Clinic vs. Video Post-Op AppointmentsTable E - SEQ Table_E_- \* ARABIC 7. Percentage of Patients that go to Surgery by Age GroupAppendix F: Patient Survey Analysis Summarized FindingsFigure F - SEQ Figure_F_-_ \* ARABIC 1. Patients' Preference to be SeenTable F - SEQ Table_F_- \* ARABIC 1. Count of Patients' Preference to be SeenFigure F - SEQ Figure_F_-_ \* ARABIC 2. Patients' Distance Traveled Compared to Access PreferencesTable F - SEQ Table_F_- \* ARABIC 2. Patients' Distance Traveled Compared to Access PreferencesFigure F - SEQ Figure_F_-_ \* ARABIC 3. Patients' Desire to be Seen Sooner Compared to Access PreferencesTable F - SEQ Table_F_- \* ARABIC 3. Patients' Desire to be Seen Sooner Compared to Access PreferencesFigure F - SEQ Figure_F_-_ \* ARABIC 4. Patients' Estimated Access Time Compared to Access PreferencesTable F - SEQ Table_F_- \* ARABIC 4. Patients' Estimated Access Time Compared to Access PreferencesFigure F - SEQ Figure_F_-_ \* ARABIC 5. Patients' PCP Involvement Compared to Access PreferencesTable F - SEQ Table_F_- \* ARABIC 5. Patients' PCP Involvement Compared to Access PreferencesAppendix G: Provider Preferences The provider preferences table is a document provided to the team that is used for scheduling purposes. From this document, the student team made a visualization that looks at each individual diagnosis and identifies which providers see that diagnosis. Figure G - SEQ Figure_G_- \* ARABIC 1. Example of Provider Diagnosis PreferencesFigure G - SEQ Figure_G_- \* ARABIC 2. Example of Provider Preference Column UpdatesAppendix H: Pugh Selection Matrix WeightingBaselineAlternative SolutionsCurrent Solution: Minimal usage of video visits for select patientsConvert varying percentage (as seen in recommendations section) of new patient appointments to e-consults and video visitsConduct 50% of post-op appointments via video visitConsider managing provider preferences from scheduling capabilityCriteriaAlignment with Michigan Medicine's virtual care task force's practices5%0550Improve patient care standard20%0322Alignment of implementation timeline with division goals5%0553Increase two-week patient access25%0323Increase overall patient access25%0433Provider acceptance10%4541Patient acceptance10%432.55Total100%0.83.652.82.65Figure H - SEQ Figure_H_- \* ARABIC 1. Pugh Selection Matrix ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download