The Use of Arena Simulation Modeling in Staffing Decisions for the Blank Child and Adolescent Psychiatry Clinic: Iowa Health Des Moines

Vanessa Louise Calderon, B.S.I.E.

Download PDF

Blank Psychiatry – Executive Summary

Blank Psychiatry Clinic is located at Iowa Lutheran Hospital in Des Moines, Iowa and includes two psychiatrists, four nurses and four administrative team members who serve as clerical staff. The clinic serves children and adolescents who suffer from anxiety, ADHD, autism, and other mental health conditions.

The Administrative Team (front desk staff) and nurses work together to prepare paper charts. To complicate this process, a new step will require the Administrative Team to process edits and denials. A new provider was set to arrive in September 2012 which would increase workloads, and pull one nurse who was fully dedicated to chart preparation away to perform patient care.

Figure 1: Strategy for Blank Psychiatry Project to Reduce Waste and Evaluate Staffing Needs.

An evaluation of Blank Psychiatry processes began in February 2012 to review processes, layouts and staffing to help improve productivity, remove waste, and reduce the need for additional FTEs.  Computer simulation models of the current and alternate future states were created to analyze the ultimate of the amount of additional staff required for chart preparation and other tasks.

A final evaluation indicated that to support patient satisfaction by maintaining patient wait times, we should budget for 0.4 FTEs of additional nursing time and 0.6 FTEs of additional administrative staff.  A cost avoidance of $42,000 / year was projected (based on original estimates of requiring 1.0 FTEs of nursing time and 1.0 FTEs of administrative staff).  By using simulation modeling we were able to recommend these increases in staffing levels with no impact in the service levels provided in this clinic.  An additional benefit of our investigations during project work included the ability to show benefits of acquiring additional space on the floor.

Blank Psychiatry – Background

Blank Psychiatry is a child psychiatry clinic located on the 3rd floor of Iowa Lutheran Hospital.  It is managed by Deb Gerlitz, RN, who also serves as the nurse for one psychiatrist [Psych #1].  Two other nurses serve Psych #2, the other psychiatrist in this department.  An additional nurse remained in the department during the initial phases of this project in anticipation of an additional physician to be added in fall 2012.  A four-person administrative team works to support this staff. Paper charts are utilized by this department in a locked room down the hall from the front desk.

Figure 2: View of Charts in Chart Room – Blank Psychiatry Clinic.

There are approximately 2,000 paper charts in a chart room down the hall from where the Administrative Team sits (creating a significant walking distance).  These charts are housed in three main cabinets that are two sided and “spin” to bring the opposite side to the front.  Staff uses GPMS (“Centricity”) to schedule patients, set up reminder calls, etc.

Blank Psychiatry serves patients aged 18 months to 18 years or the age at completion of high school (with some exceptions) who suffer from autism, ADHD, anxiety, depression, ODD, or who display disruptive or aggressive behavior.

Patient and staff satisfaction are high.  In fact, just recently, this department was awarded a special honor for achieving exemplary results in the Great Places to Work™ survey.  With this in mind, we strove to maintain patient wait times and staff utilization in order to keep these measures from regressing.

Initial Observations:  Issues Noted

Processes, Capacity, and Staffing

During our initial observations over two weeks, data was collected and interviews were conducted to capture the current state.  Front desk staff (Administrative Team) struggled to complete their tasks involving chart prepping, obtaining pre-authorizations, insurance verification, posting charges, checking in and out and scheduling patients, and processing referrals in their allotted scheduled times.  Almost all of this is related to the handling of the paper charts, faxes, and multiple locations (15-20 places observed) that a chart or contents of that chart could reside. A new process was slated to be implemented in spring of 2012 in which the Administrative Team will begin “working edits and denials.”  Staff obtained training in February 2012.

Figure 3: View of Blank Psychiatry Front Desk (Admin Team Area).

One nurse who worked for a previous physician was kept on staff in anticipation of the new physician’s [Psych #3] arrival. This nurse has been helping to “check lines” (phone messages) and prepare charts for the other doctors during this interim period without an assigned physician. Recently, the administrative team had temporary help during the week to help file charts but this is no longer available.  It is clear from these facts and the frequency of overtime for staff that they are already running near their capacity.

Press Ganey Data (Patient Satisfaction Scoring) – Blank Psychiatry Q4

Press Ganey Data [1]from Q4 2011 suggests that Blank Psychiatry does significantly better in these areas than their clinic counterparts:

Category

Other Blank Clinics

Blank Psychiatry

Moving Through your Visit

82.8

90.6

Information about Delays

83.7

93

Wait Time at Clinic

81.3

89.1

Concern of Nurse/Asst for Problem

90.6

93.8

Nurse Assistant Overall

91.7

94.1

Concern of Nurse/Asst for Problem

90.6

93.8

Table 1: Press Ganey Data – Blank Psychiatry – Q4 – Areas of Success.

The same set of data shows these as areas where Blank Psychiatry scored lower than other Blank Clinics:

Category

Other Blank Clinics

Blank Psychiatry

Access to Clinic – Overall

88.6

84.1

Ease of Getting Clinic on Phone

87.2

83.1

Convenience of Office Hours

89.1

82.3

Ease of Scheduling

87.5

82

 Table 2: Press Ganey Data – Blank Psychiatry – Q4 – Areas of Opportunity.

This data reveals that the time and effort put into preparing for the visits is paying off in the patient experience. Patients are seeing their visit as smooth and free from delays.  Still, the amount of time that goes into this process is lengthy and includes many non-value adding activities.  The project aimed to reduce the amount of administrative work to allow for more live phone calls rather than message gathering to improve other scores in this area.

Layout Issues

Figure 4 shows the floor (3W) where Blank Psychiatry is located.  Rooms in colors other than black are owned by Blank Psychiatry.  The red rooms are for providers, the purple for nurses, and the blue rooms are for administrative staff and activities (charts).  The orange rooms represent patient waiting area.

Figure 4: Layout of 3W Wing of Iowa Lutheran Hospital with Room Occupants.

The chart room is frequented by nurses and administrative staff about 10-15 times per day for pulling charts for upcoming patient visits, re-filing charts, referencing charts for medication refills, etc.  The location of the chart room increases travel time.

There has also been a desire to create a “quiet room” for patients who are particularly anxious or aggressive. The offices at the end of the hall housing non-Blank staff as well as chaplains and nurse managers represented an opportunity to rearrange spaces.

Staffing and Patient Visit Details

FTE

Physicians:

Psych #1

1

Psych #2

1

Total Physicians:

2

Nurses:

Manager, Psych #1 Nurse

1

Psych #2 Nurse #1

0.7

Psych #2 Nurse #2

0.4

Unassigned, Psych #3 Nurse

0.8

Total Nurses:

2.9

Administrative Team (Front Desk):

Admin #1

1

Admin #2

1

Admin #3

0.7

Admin #4

0.5

Total Administrative Team Members:

3.2

 Table 3: Blank Psychiatry Staff Overview.

Processes – Data Collection

Data was collected regarding the exact amount of time spent doing the following:

  • Checking in patients
  • Walking back and forth to the chart room
  • Searching for missing charts
  • Preparing charts by administrative staff and nurses
  • Processing medication refills
  • Returning phone calls, reviewing referral
  • Scheduling patients
  • Actual patient time spent with nurse/provider
  • Posting charges

Through the course of observations and shadowing, accurate times were taken in order to provide data for the later Arena models created. See Figure 7 for more detail on time distributions for these tasks.

Layout Evaluation

During short “lunch and learn” style meetings with small groups of administrative staff and nurses, different scenarios were evaluated to determine how best to utilize space. We created spaghetti diagrams to determine how additional space would benefit our processes. This was represented in the Arena models with shorter or longer travel times when demonstrating the advantages of having charts closer to administrative staff or placing someone directly in a chart room.

Arena Model Creation – Four Main Models of Blank Psychiatry Processes

What is Arena? 

Arena is a discrete event simulation and automation software developed by Systems Modeling and acquired by Rockwell Automation.

Arena Simulation allows inputs for:

  • Processes (process times with variation in a variety of distributions, i.e., normal, triangular, Erlang, Wiebel, logarithmic, etc.)
  • Resources (nurses, physicians, techs)
  • Schedules (hours worked in day, patient arrivals, phone calls per hour)
  • Entities (patients, charts, signals to do something)
  • Stations (rooms, workstations)

Model #1 – Pre-Work (Current State Findings)

The first model we created was based on the current state with the intent to validate our evaluation of processes, timings and rules for process work.  First we added in the current resources (in this case, employees) to the model:

Figure 5: Resources created based on Schedules for Pre-Work Arena Model in Blank Psychiatry.

The processes we mapped during our pre-project work were used to create workflows in our model.  We first created process maps to determine areas of focus.

Figure 6: Process Mapping for Chart Prepping Process in Blank Psychiatry (Partial).

Seven key processes were identified and mapped:

  1. “Beginning of the Day” work
  2. Refills and Referrals
  3. Chart Prepping Process (new and established)
  4. New Patient Visit
  5. Established Patient Visit
  6. Posting Charges
  7. Processing Edits and Denials

Using Arena’s “Basic Process” module, we input the time distributions for all of these processes and determined what resources (people needed, rooms, etc.) would be seized by these processes as entities (patients, charts, or signals) pass through. By time variation we mean that we enter the average and the variability (as a standard deviation) for each value.  The “Delay Type” determines what type of distribution best represents the variation in process time.  Most often, we chose to use a triangular distribution for the distributions with a “most likely” time as the center, and a high and low time represented in Figure 7 in the “Minimum,” “Value” (most likely), and “Maximum” minutes columns.

Figure 7: Basic Process Module and Resources Associated with “Check-In” Process.

Figure 7 shows that the “Check In” process seizes the resource set “s Check-In Admins” which includes Admin #1 and Admin #2, two out of the total set of four administrative members.  Whichever member is not busy will check in the patient based on the selection rule “Smallest Number Busy.”  This is a High Priority process, so this will take precedence over a lower priority process.  Direct patient interaction is always more important than other activities in our model.

One aspect of Arena simulation that allows for great power is the ability to read directly from Excel files which are easier to edit than the simulation model itself.  By inputting real schedules of patients for a typical five-day week for each physician, we can create a more realistic model.  We are also allowed to create variations in patient arrival time. Since we know a certain distribution of patients will be early and late, we created a rule that patients greater than 10 minutes late will be cancelled and a task is created for the administrative team to send out a letter to the patient explaining the “no show” and “late” policies.  Reading from an Excel file (Figure 8) also has the added benefit of allowing for easy changes to scheduled arrivals to gauge the impact on other processes.

Figure 8: Patient Arrival Schedules and Distribution of Late and Early Patient Arrivals for All Psychiatrists on All Weekdays.

 

Figure 9: Final Pre-Work Arena Model for Blank Psychiatry.

 

 

After creating all of the resources, schedules, process modules and rules, our model for pre-work was created.  A portion of the Arena Model is shown above (See Figure 9).

By running this model over 10 replications, with each replication representing a five-day week, we created output reports containing patient wait times, patient time in the system, scheduled utilization of resources, and other key indicators.  We used 10 replications for an average of the replications’ outputs.  This will be compared later in the report to the remaining three models.

Model #2 – Small Changes

During our scheduled lunch meetings with small teams of staff members, we brainstormed how best to remove additional steps and time from certain processes.  One change came in the reduction of non-recurring nurse tasks (miscellaneous work).  The simulation and subsequent discussions resulted in reduction in time for these tasks from 66 minutes to 16 minutes.  Another change was made to the chart checking process.  One administrative member would now check the charge sheets and write information on them once instead of the two times previously required.  This changed the time distribution from a triangular (45, 50, 60) minutes per day to (35, 40, 60) minutes per day.  The shift in the distribution contributed to a reduction in patient time for Psych #2 patients of approximately 10 minutes in the average patient visit time.   By making these changes and validating them with small studies, we were able to create our “post work” Arena model.

Figure 10: Reduction in Patient Time in System from Pre-Work to Post-Work (Model 1 – Model 4 in Arena) [ See headers for Model #1 – Model #4 for descriptions on what each model entails].

Model #3 – Additional Physician and No Additional Staff

The third model incorporated the addition of a physician and her proposed schedule into the model with no changes to processes or staffing.  Resource utilization was analyzed to determine how staff was strained by this additional work without any additional FTEs.  Using the International Labor Organization [ILO] model for determining fatigue allowances, a score of six points was attributed to the work of the administrative and nursing staff based on the following criteria [2] (Konz & Johnson, pp. 577-580):

  • Standing 20-39% of the time = 1 point
  • Straining 0-19% of the time = 1 point
  • Lifting weight of 20-39 lbs 0-19% of the time (patients, charts) = 2 points
  • Close application 60% of the time = 0.5 points
  • Delicate or minute work = 1.5 points

Using the ILO conversion table, we calculated a fatigue allowance of 10% that includes a 5% personal allowance (restroom breaks, refreshments, etc.)  By applying the convention that women would get a 7% allowance, we subtracted 5% from the 10% and then added back 7%.  This takes us to a 12% calculated allowance.  Due to the nature of mental health, a 5% variation allowance would be added to provide for unforeseen patient events.  In total, we agreed on an allowance of 17% which would mean that utilization of any resource over 83% in our model would be inappropriate.

Model #4 – Additional Physician and Changes to Staff (Utilization Reduction)

After studying the effects of the additional provider on utilization, different scenarios were demonstrated using the scheduling module of Arena.  Two resources were created: “r Extra Nurse” and “r Extra Admin.”  By adjusting their schedules and time in the office in the computer simulation, we were able to find a solution that reduced and smoothed out utilization of all resources in the system to below our 83% maximum utilization threshold (Figure 11), but also maintained or reduced patient time in the system (our balancing measure) for each provider (Figure 10).  The final solution was to add 0.4 FTE of an additional nurse (to be added to the schedule of a current nurse) and 0.6 FTE of additional administrative support staff.

Figure 11: Resources and Scheduled Utilization Percentages for Each Arena Model (Models 1-4).

Final Decisions Based on Modeling:  Cost Avoidance and Layout Changes

By creating our Arena models to aid in decision making, we not only created concrete reasons to add staff in a challenging financial climate, but we also were able to avoid over-budgeting for staff that would not be necessary to maintain the patient experience in this department.  Figure 10 gives the reduction in patient visit time for the original two doctors (Psych #1 and Psych #2).  The scenario in model 1 gives a visit time of about 48.04 minutes for Psych #2 in current state vs. 35.95 minutes in scenario 4 (future state with added staff for a reduction of over 10 minutes).  The change in Psych #1 patient time is negligible (33.27 in scenario 1 to 33.39 in scenario 4).  This non-change is acceptable as it is not statistically significantly higher.

Using conservative estimates, the savings of avoiding 0.6 FTE of nursing staff and 0.4 FTE of administrative staff equates to about $42,000 in annual savings (See Table 4; $30,000 + $12,000 = $42,000 Dollar ($) Savings).

Job Role

Estimated Need Before Work

Estimated Need After Work

Dollar ($) Savings

Nurses

1.0 FTE

0.4 FTE

(1.0-0.4)*$50,000[3] = $30,000

Administrative Staff

1.0 FTE

0.6 FTE

(1.0-0.6)*$30,000[4] = $12,000

Table 4: Estimated Additional Staffing Needs Before and After Process Improvement Work and Analysis for Blank Psychiatry.

We were also able to demonstrate the benefits of acquiring additional floor space.  By working with administration and construction, we created a plan to shift offices so that Blank Psychiatry will now have the room behind the administrative office for their charts, as well as another space for a quiet room for patients.

Additional Benefits of Using Simulation Modeling

The time commitment in conducting this project could be weighed against this cost for the hourly rate of the process improvement work; however, the strategic benefit of creating a culture of improvement far outweighs the small hourly commitments given to this project.  In addition, the creation of these Arena simulation models can and likely will be used by these areas again as they are built to be easily modified to simulate changes to processes in the future.

Works Cited

Konz, S., & Johnson, S. (n.d.). Work Design: Industrial Ergonomics. 2000: Holcomb Hathaway, Inc.

Vanessa Louise Calderon, B.S.I.E., is a Process Improvement Coordinator for Iowa Health – Des Moines. She was raised in Des Moines, IA and earned her I.M.S.E. degree from Iowa State University’s College of Engineering.  Her current interests include computer-based simulation modeling, and facilitating teams in meeting compliance and financial goals in a healthcare setting. 1200 Pleasant St. Des Moines, IA 50309,  (515) 241-6887  Email:  [email protected]

Acknowledgements

Financial Advocates (Administrative Team):  Carla Watts, Laurie Frank, Ronda Knight, Mary Jane Mehre

Nurses:  Dawn Angell, RN; Erin Boyd, RN; Teresa Hensley, RN; Colleen Milburn, RN; Deb Gerlitz, RNC, BS

Psychiatrists:  Dr. Kevin Took, Dr. Leenu Mishra

 


[1] Press Ganey Data is a measure (out of a possible 100 points) of patient satisfaction on a number of categories.  Surveys completed by patients are returned to and compiled by the Press Ganey Company for use by each department.

[2] Konz, S., & Johnson, S. (n.d.). Work Design: Industrial Ergonomics. 2000: Holcomb Hathaway, Inc.

[3] Estimate of Nursing Salary based on internal company standards.

[4] Estimate of Administrative Staff Salary based on internal company standards.