background targeted changes analytical approach optimization
Transcription
background targeted changes analytical approach optimization
Using Simulation Based Optimization To Structure And Manage Ambulatory Clinics Phil Troy, Nadia Lahrichi, Dana Porubska and Lawrence Rosenberg BACKGROUND · The Sir Mortimer B. Davis Jewish General Hospital · A full service university affiliated medical center a broad range of inpatient and outpatient · Provides services major tertiary and quaternary cardiovascular, · Has neurosciences, oncology (including robotic surgery) · · · TARGETED CHANGES and colo-rectal programmes Performs 13,000-15,000 operative procedures per year To improve patient outcomes and satisfaction the hospital decided to standardize and expand the services provided by its Pre-Admission Testing (PAT) Clinic · Patient Sees Surgeon In Surgeon's Office Patient Goes Through Screening Process Patient Undergoes Procedure At Hospital · Patient Recovers Of JGH's Pre-Admission Testing Clinic · Activities Patient logistical preparations (phone call) · · When to arrive to the clinic · Expected wait time at the clinic · Information and preparation for the visit · Administrative (on-site) · Register · Submit insurance information · Medical (on-site) · Obtain medical/surgical history · Perform physical exam · Adjust medications · Take ECG · Perform other tests as required · Refer abnormal results to surgeon for follow-up · Clinic Issues · Existing Not enough space: · · · · · · · · · · Waiting area (wheelchair access) · Exam rooms · Individual training rooms · Administrative work area Not enough nurses and administrative technicians Teaching provided only to same day surgery patients Inefficient patient flow resulting in long wait times Inconsistent medication adjustments Inconsistent handling of abnormal results Not completing charts on-time Inadequate follow-up Lack of comprehensive services increases post-op complications (literature) · Provide medication reconciliation for all patients · Provide training for all patients · Increase nursing and administrative staffing · Streamline patient flow · Identify patients with infections and allergies · Consistently manage abnormal results · Complete charts at least 72 hours before procedure Pre-Surgical Screening Clinic Activities · Proposed Up to 35 patients/day would do some of the following: A Simplified View Of JGH's Peri-Operative Process Patient Is Referred To Surgeon By GP Analysis of the Existing Clinic Indicated Need To: · Register for the clinic · Submit insurance information · Watch a video orientation at the start of visit · See a pharmacist · Change into a gown · Have ECG taken · See General Practitioner (GP) · See Internist · Get dressed · Provide blood and urine samples · Receive group or DVD training · Receive individual training · Undergo X-rays test Management Challenges · Minimizing/determining space requirements staffing costs (including · Minimizing/determining overtime) · Minimizing/preventing physician idleness · Minimizing excessive patient waiting · Handling patients having differing needs · Handling uncertainty about times needed for each task · Handling patients' need to see pharmacist · Handling no shows and cancellations · Making sure that staff have breaks and lunch staff and patients while taking into · Scheduling consideration the above challenges REFERENCES discrete event simulation model to evaluate · Aoperational performance of a colonoscopy suite, Bjorn Berg, Brian Denton, Heidi Nelson, Hari Balasubramanian, Ahmed Rahman, Angela Bailey, Keith Lindor, Medical Decision Making (2009), 3, 380-387. · Reducing patient wait times and improving resource utilization at British Columbia Cancer Agency's ambulatory care unit through simulation, Santibáñez P, Chow VS, French J, Puterman ML, Tyldesley S., Health Care Management Science (2009), 12, 392-407. ANALYTICAL APPROACH · SIMULATION ANIMATION SNAPSHOT Simulation Based Optimization RESULTS AND CONCLUSIONS ● · To handle processing time variability · To facilitate the development of acceptable model · To facilitate measurement of multiple statistics · To facilitate what-if analysis · To facilitate automatic scheduling improvements Staff And Patient Schedule Entity Type Admissions Staff Type Count Times Arrival 2 6:00 6:00 Break 8:55 8:55 Lunch 11:50 11:50 Registered Nurse Arrival 3 7:00 7:00 7:05 Break 9:55 9:55 9:50 Lunch 12:25 12:25 12:20 Pharmacist Arrival 1 7:30 Break 9:30 Lunch 12:00 ECG Technician Arrival 1 6:25 Break 8:25 Lunch 10:55 Blood Taker Arrival 1 7:55 Break 10:05 Lunch 12:40 General Practitioner Arrival 2 6:30 7:00 Internist Arrival 1 10:10 DVD Player 12 Patient Arrival 35 6:00 6:00 6:00 6:00 6:00 6:00 6:00 6:05 6:05 6:05 6:05 6:05 6:20 6:25 6:35 6:45 7:30 7:30 7:40 7:40 7:50 8:10 8:15 8:20 8:20 8:25 8:40 8:50 9:00 9:05 9:30 9:45 10:05 10:35 11:40 Simulation Model · TheChallenges · entities so as to allow dynamic and · Modelling individual decision making activities involving different entities that · Modelling needed to be coordinated · Simulation modelling approach · Could have treated: ·Patients as entities ·Staff, exam rooms, DVD players, . . . as resources · Instead modelled all objects as entities · Animated state information for each entity type · Used logic to handle events and entity flows · Requirements · Tasks needed for each patient profile · Patient profile distribution · Service time distributions · Count of tasks needing to be done each day · Actual simulation model data profiles guesstimates from Subject Matter · Patient Experts (SMEs) in existing (PAT) clinic · Service time distributions ·Triangular distribution guesstimates from SMEs ·Patient self-time studies ·Count of tasks needing to be done each day · Room allocation issues · Determine room allocation by physician type · Preclude room pooling between GP and Internist · Internist must have two rooms · Validating the simulation model · Was difficult · Plan for Pre-Surgical Screening (PSS) clinic is in flux · Incomplete data · Received feedback from management ·Pre-Surgical Screening Clinic Nursing Coordinator ·The Chief Of Surgical Services ·Associate Director Of Professional Services against schedule with deterministic service · Tested times were sensitive to values of service time · Results distribution guesstimates ● Entity Type Admissions Staff Registered Nurse Pharmacist ECG Technician Blood Taker General Practitioner Internist Patient OPTIMIZATION · Optimization Problem · Objective function - minimize sum of costs of: Physician idle time Staff overtime Excessive patient waiting time Constraints Finish all patient tasks by end of work day Respect staff break and lunch times Both general practitioners work in the morning The sole internist works in the afternoon At least 8 people in group training sessions Sending staff home at end of their shift if there is another staff member who can finish up for them · · · · · · · · · · · Optimization Issues And Challenges · · · · Need to finish all patient tasks before end of work day Certain staff had to arrive before other staff Breaks and lunches had to fit into 8 hour day Not enough nursing staff · Optimization Procedure · · Initially used Optquest to minimize physician idle time Subsequently developed simple search heuristic Set initial values for each time variable Set upper and lower bounds for each time variable Select initial time increment Repeat forever Loop through variables one at a time Change variable positively and then negatively by time increment If solution is feasible evaluate average of total simulated cost over a pre-determined number of days Keep test solution if it is an improvement Gradually decrease magnitude of time increment · · · · Overtime, Idle Time And Excessive Waiting Costs · · · · · Source Of Cost $/Hour Overtime $30 Overtime $80 Overtime $80 Overtime $45 Overtime $45 Idle time $100 Idle time $100 Excessive waiting $20 Mean Total Hours/Day Cost/Day 0.15 $4.50 0.13 $10.67 0.00 0.00 0.00 0.18 $18.33 0.02 $1.67 2.65 $53.00 $88.17 Simulation Had No Method To Substantiate · Before Amount of physician idle time · · Amount of staff overtime · Number of exam rooms · Staff requirements A Result · AsPhysicians believed they would have a lot of idle time · · · Physicians pushed to each have three exam rooms · Professional services pushed for less rooms · No realistic understanding of staff requirements · No understanding of effect of breaks on throughput Using Simulation With Different Assumptions · Was able to determine room requirements · Was able to quantify expected physician idle time · Was able to quantify expected overtime able to identify bottleneck that would have very · Was significantly increased excessive patient waiting times able to improve scheduling for physicians, staff · Was and patients