Michigan Medicine Quality Improvement
Continuity of Care in Stroke Patient Discharge
Current State Analysis and Recommendations
December 12, 2017
To:
Jenevra Foley, Operations Director, Comprehensive Stroke Center,
Alison Spork, Senior Project Manager, Physical Medicine & Rehabilitation,
Mark Van Oyen, IOE 481 Professor,
Mary Duck, Industrial Engineering Expert,
From:
IOE 481 Project Team #8
Christopher Cardoso, IOE 481 Student,
Joshua Johnston, IOE 481 Student,
Catherine Samuel, IOE 481 Student,
Table of Contents
EXECUTIVE SUMMARY
Background
Key Issues
Methodology
Interviews
Data Collection
Data Analysis
Summary Statistics
Current State Flow Map
Findings and Conclusions
Opportunity to Increase Transparency
Some Aspects of the Process Are Out of the Hospital's Control
Data Findings for Waiting Times in Process
Need for Additional Data and Consistency in Data Collection
Recommendations
Future Recommendations
INTRODUCTION
BACKGROUND
KEY ISSUES
GOALS AND OBJECTIVES
PROJECT SCOPE
DATA AND METHODS
Design Methods, Constraints, and Standards
Data Collection
Literature Search
Interviews
Data Analysis
Current State Flow Process Maps with Data
FINDINGS AND CONCLUSIONS
Opportunity to Increase Transparency
Some Aspects of the Process Are Out of the Hospital's Control
Quantitative analysis of Discharge/Admission Process Timing Shows High Variability
Opportunities for Streamlined Data Tracking
Findings from Literature Provided Context
RECOMMENDATIONS
Immediate Recommendations
1. Implement Excel data collection tool for use by stakeholders
2. Use metrics to benchmark process timing and improve performance
3. Improve communication with pages at key decision points.
Future Recommendations
1. Create additional Excel tool that can pull data from MiChart reports
2. Use Time Studies to validate in process times.
3. Use additional data collected to complete a full six-sigma analysis
EXPECTED IMPACT
REFERENCES
APPENDIX A: Design Documentation
Constraints
Standards
APPENDIX B: Current State Process Map and A3 Documentation
APPENDIX C: Results from Data Analysis
APPENDIX D: Work Instructions for Excel Tool35
List of Tables and Figures
Table 1: Summary Statistics of Individual Steps in Discharge/Admission Process / 8Table 2: Median and Standard Deviation for Highly Variable Process Data Points / 10
Table A-1: Constraints and Standards Matrix / 18
Figure 1: Distribution of timing from Stroke Admission to IRF Admission / 11
Figure B-1: 4A to IRF Admission Process Flowchart using data without outliers / 20
Figure B-2: Orders/Consults Current State Process Flowchart using data without outliers / 21
Figure B-3: Insurance Approval Current State Process Flowchart using data without outliers / 22
Figure B-4: Transfer from 4A to IRF Current State Process Map with Data using data without outliers / 23
Figure B-5: Key to the 4A to IRF Admission Process Flowchart / 24
Figure B-6: Project A3 Document / 25
Figure C-1: Distribution of Timing from Stroke Admission to PT/OT Order / 26
Figure C-2: Distribution of Timing from Stroke Admission to PT/OT Order with Outliers / 26
Figure C-3: Distribution of Timing from PT/OT Order to PT Note / 27
Figure C-4: Distribution of Timing from PT/OT Order to PT Note with Outliers / 27
Figure C-5: Distribution of Timing from PT/OT Order to OT Note / 28
Figure C-6: Distribution of Timing from PT/OT Order to OT Note with Outliers / 28
Figure C-7: Distribution of Timing from PT Note to PM&R Order / 29
Figure C-8: Distribution of Timing from OT Note to PM&R Order / 29
Figure C-9: Distribution of Timing from PM&R Consult to PM&R Note / 30
Figure C-10: Distribution of Timing from PM&R Consult to PM&R Note with Outliers / 30
Figure C-11: Distribution of Timing from Being Added to Potential Admit List to Insurance / 31
Figure C-12: Distribution of Timing from Insurance Verification to Insurance Approval / 31
Figure C-13: Distribution of Timing from PM&R Note to Discharge Order / 32
Figure C-14: Distribution of Timing from PM&R Note to Discharge Order with Outliers / 32
Figure C-15: Distribution of Timing from 4A Discharge Order to Actual 4A Discharge / 33
Figure C-16: Distribution of Timing from 4A Discharge Order to Actual 4A Discharge with Outliers / 33
Figure C-17: Distribution of Timing from Actual 4A Discharge to 6A Admission / 34
Figure D-1: Excel Tool Work Instructions for Order Sheet / 35
Figure D-2: Excel Tool Work Instructions for Summary Sheet / 36
1
EXECUTIVE SUMMARY
Michigan Medicine’s Comprehensive Stroke Center treats approximately 700 stroke patients every year [1]. A vast majority of these stroke patients are housed in the Neurology unit (4A). A subset of the stroke population in 4A moves to the Inpatient Rehabilitation Facility (IRF) after their time in 4A is complete, and there is currently a perception that this process takes too long. Therefore, the Comprehensive Stroke Center and the IRF have requested that an IOE 481 student team analyze the process to identify bottlenecks and develop metrics that the units can use to track performance. Through interviews with stakeholders and data analysis regarding timestamp data, the student team created a current state flow map with recommendations for increased communication. Lastly, the team has created recommendations for future analysis and data collection.
Background
This project is a continuation of work that was completed by another IOE 481 student team in the Winter of 2017. The previous students analyzed the current state of the process at a high-level in order to gain an understanding of where future studies should focus their efforts [2]. With an average of ~89% capacity in the stroke unit and 90+% occupancy in the IRF [3], there is pressure on both ends of the process to manage throughput as efficiently as possible. For this project, the student team has taken a deeper look at the discharge/admission process to make recommendations for improvement.
Key Issues
The following key issues are driving the need for this project:
●Discharge/admission process between 4A and IRF is complex and not readily visual
●Data collection is not currently a standardized process
Methodology
The student team used the following methods throughout the project.
Interviews
The team conducted 8 interviews of key stakeholders, including the IRF Admissions Coordinator, the 4A Case Manager, a doctor in the Stroke Unit, a Speech Language Pathologist, a Physical Therapist, a doctor in the IRF, a Stroke Unit nurse, and the Insurance Coordinator in the IRF. These interviews were integral for developing a thorough understanding of the current state of the discharge/admission process and helped the team begin to understand the current state of how units communicate with each other.
Data Collection
The team used two major data collection sources. The first source is from the IRF admissions coordinator, which includes timestamp information for 20 patients between February 2017 - October 2017 regarding each step in the IRF admissions process. The second collection source was manually pulled via MiChart, where the team collected timestamp data for 40 patients between January 2017 - November 2017 regarding each step in the 4A discharge process.
Data Analysis
The team analyzed the collected data in order to generate summary statistics and create a process flow diagram.
Summary Statistics
Information from both interviews and data collection in MiChart was used to generate summary statistics for the steps within the discharge/admission process. Using timestamps from the collected data, the team calculated the time between process steps, and used this to generate summary statistics regarding the mean, median, standard deviation, 80th percentile, minimum, and maximum, as well as create box plots to better understand the distributions. These figures can be found in Appendix C.
Current State Flow Map
Using the interviews from eight key stakeholders in the process, the team generated a current state process flow map of the discharge/admission process. The map separates the process into three major sections: Orders/Consults, Insurance Verification and Approval, and Admission to IRF. The whole process map and each individual section can be seen in Appendix B. All available data used to generate summary statistics was added to the current state flow map to better quantify the length of the process. For more information, the current state flow map can be found in Appendix B.
Findings and Conclusions
Using the methods outlined above, the team was able to draw the following conclusions.
Opportunity to Increase Transparency
Through interviewing stakeholders, it became evident that stakeholders were unaware of the complex steps involved in the admission/discharge process, and that there are opportunities to improve communication between the 4A Case Manager and the IRF Utilization Review Coordinator. The team worked to mitigate these issues by creating a Current State Flow Map, which will help stakeholders better understand the overall process, as well as recommendations for standardized communication between unit 4A and the IRF, to reduce communication barriers between units. The map can be found in Appendix B.
Some Aspects of the Process Are Out of the Hospital's Control
Two of the largest barriers for a smooth and timely transition from 4A to the IRF are insurance approval and high bed demand. There are 32 beds available and the hospital is not able to control the strong demand. These are the major bottlenecks that were identified in the discharge/admission process.
Data Findings for Waiting Times in Process
With the data that the student team was able to extract from MiChart, many different metrics were developed. These metrics were then analyzed using statistical analysis to determine a quantitative average state. With this data the team created box plots to show the spread and variation of the data. The team also created an Excel tool that will standardize data collection by automatically generating timestamps. The 4A Case Manager, the IRF Admissions Coordinator, and the IRF Utilization Review Coordinator will no longer use manual data entry, which will standardize format and therefore assist stakeholders in future reporting and analysis.
Need for Additional Data and Consistency in Data Collection
In order to conduct a more thorough analysis of the process, more data will be required. While the team was able to pull enough data points for each statistic to meet hospital standards for statistical significance, much of the data that the team was presented with was either invalid, incomplete, or considered an extreme outlier when compared to the other values.
Recommendations
The team has made the following recommendations for improving the discharge/admission process.
Immediate Recommendations
1. Implement Excel data collection tool to assist 4A Case Manager, IRF Admissions Coordinator, and IRF Utilization Review Coordinator with data collection and analysis
2. Use determined data metrics to create benchmarks for process timing that stakeholders can use to improve performance
3. Improve communication by sending pages at key decision points to the 4A Case Manager denoted on the current state process map in Appendix B
Future Recommendations
1. Create additional Excel tool that can pull data from MiChart reports to remove the need for manual data collection
2. Conduct Time Studies to validate the in-process times for each step
3. Use additional data collected to complete a full six-sigma analysis on the process
INTRODUCTION
Michigan Medicine’s Comprehensive Stroke Center treats approximately 700 stroke patients every year [1], many of which begin their stay at the hospital in the Neurology Unit (4A). After stroke patients receive the care they need, there are a variety of pathways for patients to continue their care, including admission to Michigan Medicine’s Inpatient Rehabilitation Facility (IRF). Currently, there is a perception that patients are waiting in 4A longer than necessary while tasks that allow their discharge to the IRF are being completed. This perception is due to a disconnect in communication and knowledge of the process among stakeholders. There is also limited data regarding the time it takes to perform the activities required before a patient can be discharged from 4A or admitted to the IRF. Consequently, the 4A and IRF units requested a team of IOE 481 students to analyze the process of discharge from 4A to the admission into IRF. This information was then used to identify and document the bottlenecks in the patient discharge process, as well as develop metrics that the units can use to track their performance. The team has also created a tool to help stakeholders track the amount of time they take in a uniform way and made recommendations to increase communication in the future. The purpose of this report is to document the work completed by the team as well as the expected outcomes of that work.
BACKGROUND
This project is a continuation of work that was completed by another IOE 481 student team in the Winter of 2017. The previous students generated a high-level process map and A3 documents in order to gain an understanding of where future studies should focus their efforts [2]. For this project, the student team will be focusing on discharge from 4A to an admission to the IRF on a much deeper-level. This newfound information will be added to the previous project’s existing
A3 document. After a patient is deemed appropriate for acute rehabilitation by the Physical Medicine and Rehabilitation (PM&R) consult physician, several steps must occur before that patient can be discharged and admitted to the IRF. First, the 4A case manager must be notified of the patient’s planned disposition so that appropriate discharge planning can take place. The IRF Admissions Coordinator must also be notified to initiate the admissions process, which involved several sequential processes. For example, proper insurance benefits must be verified, and the admission must be authorized by the insurer, which requires that specific documentation from the patient's stay on 4A is complete and available. Currently there is a perception of breakdowns in the communication between 4A and the IRF, and unnecessary delays. This leads to stakeholders believing that patients staying in 4A for extended periods of time while they wait for their discharge to the IRF. With an average of ~89% capacity filled in the stroke unit and 90+% occupancy in the IRF [3], there is pressure on both ends of the process to manage throughput as efficiently as possible. Currently there are no standard metrics that are tracked in this process to determine current performance. In addition, there are no tools to assist in data tracking and analysis.
KEY ISSUES
The following key issues are driving the need for this project:
●Inadequate levels of transparency in the discharge/admission process between 4A and the IRF
○There is a perception that the process takes too long
○Current communication practices leave some stakeholders with insufficient information
●Opportunities for standardized data collection
○Data is being collected for most of the steps in the process, but it is manually inputted and there is an opportunity for human error
GOALS AND OBJECTIVES
In order to best increase transparency regarding this process, the team was tasked with the following goals:
●Interview stakeholders involved in the process of discharging a patient from 4A and admitting to the IRF
●Analyze collected timestamp data for the discharge/admission process
With this collected information, the team delivered the following objectives:
●Recommended trackable quality metrics
○Metrics will be selected by prioritizing their urgency based on high variability and importance to stakeholders
●Developed an Excel tool that standardizes data collection
●Created a detailed current state process map that lays out each step in the discharge/admission process to improve transparency
○Data will be added to this where possible
●Added details to existing A3 diagram that documents the current state of the discharge
process
●Generated recommendations for future communication between stakeholders in the process
●Identify bottlenecks in the discharge/admission process that may be causing unnecessarily long wait times
PROJECT SCOPE
Because of the complexity of this process, there was a need to define a project scope as shown below.
In Scope
The project scope includes the time at which a patient is admitted to 4A with a stroke until the patient arrives in an IRF bed.
Out of Scope
Any clinical decision or treatment planning completed in 4A and IRF are excluded from the scope of this project. In addition, any patients who get discharged from any other Michigan Medicine unit and home is outside of scope. Lastly, this project is focused only on ischemic stroke patients.
DATA AND METHODS
In order to complete the goals of the project, the team has worked to analyze the qualitative and quantitative information that has been provided through interviews and from MiChart data. The analysis has been used to create the process flow map that will be used in to increase transparency of the current process.
Design Methods, Constraints, and Standards
A primary deliverable of the project is to give the stakeholders a better idea of the process a patient goes through from 4A to the IRF unit. The team has achieved this through the use of the engineering design process. The constraints and standards are located in Appendix A - Design Documentation to further explain the team’s key considerations when designing our final deliverables.
Data Collection
To determine timing information between each step in the process, the team concluded that data collected directly from the hospital is the best method. Completing time studies to collect this information proved to be an inferior method because the process happens sporadically, and the team wished to use previous data in order to gather large enough sample sizes for a robust data analysis. Therefore, the team used two main avenues for data collection: information collected by the IRF Utilization Review Coordinator and MiChart timestamps. The information provided by the IRF Utilization Review Coordinator included timestamps for when patients were placed on the potential admit list, insurance verification, insurance approval, able to admit notification, and actual admit time. The data provided on this set was directly related to the insurance verification and approval as well as actual admission to the IRF. The data however did not explain much about the orders and consults side of the discharge/admission process. Therefore, timestamp data was pulled either manually in MiChart or through MiChart Reports for physical therapist (PT) and occupational therapist (OT) order, PM&R order, PT Note, OT note, PM&R Note, discharge order, and official discharge time.