S746
International Journal of Radiation Oncology Biology Physics
1 failure mode maintained RPN significance in the post-FMEA workflow: patient motion after alignment or during treatment. Conclusions: Performing a FMEA before clinical implementation has significantly strengthened the safety and feasibility of our new workflow. The FMEA proved a valuable evaluation tool, identifying potential problem areas so that we could create a safer workflow. Author Disclosure: R.T. Jones: None. L. Handsfield: E. Research Grant; research supported by Funding Opportunity Number CMS-1C1-12-0001 from Centers for Medicare and Medicaid Services, Center for Medicare and Medicaid Innovation. P.W. Read: E. Research Grant; research supported by Funding Opportunity Number CMS-1C1-12-0001 from Centers for Medicare and Medicaid Services, Center for Medicare and Medicaid Innovation. D.D. Wilson: None. R. Van Ausdal: None. D.J. Schlesinger: None. Q. Chen: None.
3360
3359 Management of Task Saturation in a Pancreatic Multidisciplinary Clinic: Using a Novel Operations Management Methodology to Increase Efficiency and Reduce Unnecessary Downstream Encounters S. Moningi, S.M. Elnahal, A. Wild, A.S. Dholakia, M. Hodgin, P. Huang, and J.M. Herman; Johns Hopkins University School of Medicine, Baltimore, MD Purpose/Objective(s): Multidisciplinary clinics (MDCs) are associated with higher patient satisfaction and superior clinical outcomes for patients with pancreatic cancer. However, concerns about efficiency and patient volume hinder the model’s wider adoption. Coordination challenges often cause providers to reach a state of “cognitive overload,” when they are burdened with too many tasks within a short timeframe. Missed tasks lead to inefficiency and preventable problems. The Military Acuity Model (MAM) is an operations management methodology that redistributes high-value tasks to ensure more reliable task completion. After one year of implementation in a pancreatic MDC (PMDC), we aimed to demonstrate a significant increase in patient volume, all while keeping constant resources and reducing preventable downstream clinical encounters. Materials/Methods: MAM approached operations redesign in our MDC with 3 main strategies: process mining (elucidating required tasks for each clinical objective); cognitive load assessment (determining the number of tasks at which each provider will begin missing them); and process arbitrage (redistribution of high- value tasks). A pre-clinic questionnaire was used to gather information and assign tasks to specific staff before each clinic day. Resulting interventions were employed by an existing clinic manager who (1) assigned an appropriate provider for each patient based on disease stage and specialty (2) reconciled pre-clinical imaging to ensure up-to-date staging information and to initiate triage for emergencies (3) moved pain assessment tasks earlier in the day to ensure adequate evaluation. Patient charts from 2012 to 2014 were evaluated for daily patient volume, volume per room, and downstream phone calls to clinic and ED visits. Pre and post-implementation outcomes were compared using Fischer’s exact test and two-sided t tests. Results: A total of 91 patients were seen from June-December 2012 at the JHH PMDC, before MAM was implemented. Patients seen post-implementation (2013-2014) were 324. Patients per room per day increased from mean 1.6 to 2.6, an increase of 57.3% (p<0.001). Physician room time was cut from approximately 40 to 20 minutes per patient. Patients requiring phone calls back to clinic with unresolved issues decreased from 34 to 22% (p Z 0.049). Patients requiring post-clinic ED visits decreased from 9.9 to 7.9% (NS). Conclusions: MAM increased productivity with less resources deployed (physician time and room space), all while decreasing the number of unnecessary downstream encounters in an MDC. Its use should be validated for comparable value enhancement in other clinic settings. Author Disclosure: S. Moningi: None. S.M. Elnahal: None. A. Wild: None. A.S. Dholakia: None. M. Hodgin: None. P. Huang: None. J.M. Herman: None.
A 3D Patient-Specific Collision Avoidance System for Radiation Therapy Treatment Planning K. Fritscher,1 M. Mehrwald,2 P. Steininger,2 F. Sedlmayer,3 and H. Deutschmann4; 1Medphoton, Salzburg, Austria, 2Paracelsus Medical University, Medphoton, Salzburg, Austria, 3Salzburger Landeskliniken, Paracelsus Medical University, Salzburg, Austria, 4 Salzburger Landeskliniken, Paracelsus Medical University, Medphoton, Salzburg, Austria Purpose/Objective(s): Because of the increasing complexity and large variety of treatment scenarios in radiation therapy, systems which prevent collisions between patient, patient positioning systems, gantry or imaging devices are of utmost importance for the safety of patients, personnel and machine parts. However, most commercial treatment planning systems offer little or no possibilities for collision prevention. Consequently, a system that is capable of performing off-line virtual simulation of patientspecific beam plans during treatment planning as well as on-line collision avoidance during control of movements in the irradiation room has been developed. Materials/Methods: Collision avoidance is based on a detailed room specific 3D description of static objects in the treatment room (e.g. couch, collimator or nozzles) and dynamic objects, like patients. In the absence of a 3D surface scanner in the treatment room, the collision avoidance system (CAS) provides the possibility to create, which can act as a surrogate for patient surface scans or whole-body-outlines derived from 3D image volumes. Using an expandable set of patient specific body measurements, a fully textured 3D body surface representation based upon an anatomically correct musculoskeletal model is created. The pose of the avatar is automatically adapted according to the prescribed treatment plan. Static as well as dynamic objects serve as an input for the CAS together with treatment plan prescriptions of which a sequence of room control points, which represent discretized motion trajectories of all movable axes in an irradiation room, is extracted. In accordance with specific uncertainties concerning the exact position and/or shape especially of dynamic objects, isotropic safety margins can be defined for each object. Results: The presented system, implemented as a standalone software tool, was tested with different room and patient configurations including highly complex setups with a ceiling mounted 6-axis robotic patient positioning system. All potential collisions have been detected correctly and reliably. All calculations are performed on the CPU. Using a current quad-core processor, the average calculation time for a collision check per object pair is w2.25 ms. As a result, the system can be used for off-line treatment planning, but also for on-line collision avoidance. Conclusions: A reliable and intuitive system for patient and room specific collision avoidance has been developed. All relevant objects within a treatment room as well as patients and patient fixation devices are modeled in detail in order to provide highly accurate on- and off-line collision detection. By this means the safety of personnel, patients and equipment can be significantly improved and time consuming re-planning due to potential collisions can be avoided. Author Disclosure: K. Fritscher: None. M. Mehrwald: None. P. Steininger: None. F. Sedlmayer: None. H. Deutschmann: None.
3361 Designing a Patient Treatment Workflow Management and Analysis System in a Department of Radiation Oncology C. Liu, A.R. Yeung, J. Greenwalt, K. Mittauer, S. Samant, and R.A. Zlotecki; Department of Radiation Oncology, University of Florida, Gainesville, FL Purpose/Objective(s): To monitor and improve treatment workflow with a checklist-based software strategy and a “good catch” program. Materials/Methods: Numerous technical tasks are required prior to and during radiation therapy to confirm that patients receive accurate and safe treatment. Radiation oncology departments must ensure efficient workflow to treat a diagnostically diverse population while adhering to
Volume 90 Number 1S Supplement 2014 stringent regulatory and billing requirements. Thus, we developed and implemented (Dec 2013) software written in VB.NET and using serial workflow based on a checklist philosophy used in vertically integrated manufacturing. The software identified and tracked completion of tasks appropriate to each patient’s treatment, including generation of documentation and multiple/parallel monitoring points. This software was integrated with MOSIAQ, a common electronic medical system, to autodeposit the generated documents that indicate the listing status of required tasks for each staff member. Microsoft Outlook API was used to communicate between staff and coordinate resolution of issues through email or text messaging. A “good catch” program was designed to identify the workflow incidents and technical errors in each step, enabling the department quality control and improvement team to alter existing processes or implement new workflows. The software included time analysis functionality to analyze the completion times for tasks and “good catch” analysis to identify not only common problems in each group but also “good catch” staff. Results: This checklist system was successfully implemented in our department, and clinical/ communication issues were identified and improved. During the implementation of our first version, additional checklists were identified to improve patient treatment safety issues. In addition, a patient contrast form was integrated into the program to improve clinical operation. Following implementation, we found that the system (1) reduced regulatory and treatment documentation compliance events, (2) identified communication problems, (3) empowered staff to submit “good catch” issues for the department to improve workflow and treatment quality and safety, and (4) enabled the department, through the time analysis program, to easily identify and improve system bottle necks. Conclusions: The checklist-based software successfully monitored clinical workflow efficiency, safety checks, and regulatory compliance for individual patients and treatment populations. The “good catch” program and time analysis helped to identify recurring safety issues and workflow inefficiencies for further review by the quality control and improvement team. Author Disclosure: C. Liu: None. A.R. Yeung: None. J. Greenwalt: None. K. Mittauer: None. S. Samant: None. R.A. Zlotecki: None.
3362 Determination of Risk-Based Target Margins: A Failure Mode and Effect Analysis (FMEA) Proof of Principle B. Yi, W. D’Souza, and K.L. Prado; University of Maryland, Baltimore, MD Purpose/Objective(s): Methods for determining margins have been suggested that are based on statistical localization distributions. Few methods provide a systematic way to determine margins customized to a clinical situation. Due to this lack of a proper system, decisions are often made ‘clinically.’ Failure Modes and Effects Analysis (FMEA) is a systematic method for evaluating a process to identify how it might fail and to assess the impact of failures. This concept could be applied to determining a riskbased customized margin. The purpose of this study is to develop this method to determine margins dependent upon clinical and operational situations. Materials/Methods: In FMEA, the risk priority number (RPN) is defined as a product of the three quantities: occurrence (O), severity (S) and probability detection (D). When margins are not sufficient, the probability (occurrence O) of missing the target will be increased. Detectability (D) is used here as a means to adjust the target position. Since target coverage is associated with clinical outcome, two levels of compromised target coverage (clinical tolerance) are introduced: Compromised Clinical Intention (CCI) -1 and CCI -2. CCI-1 accepts an additional 1 % loss of the population tumor control probability or TCPpop and CCI-2 accepts another 3% loss, in order to allow a reduction in the margin. Severity decreases for CCI. A reference margin (RM) is defined as a margin for standard conditions: all O, S and D Z 5. The RM can be determined by one of many methods. In our study, the RM is calculated using the
Poster Viewing Abstracts S747 Scientific Abstract 3362; Table Examples of Individualized Risk Based Margins Margin determination using FMEA Clinical Tolerance Standard Standard Standard
CCI-1 CCI-1 CCI-2
Margin RM 20% smaller than RM 20% smaller than RM 20% smaller than RM 20% smaller than RM 20% smaller than RM
O
S
D
RPN
Evaluaton
Standard Standard
Extra efforts
5 6
5 5
5 5
125 150
Acceptable >125
Increased Localization Precision Standard
6
5
4
120
Acceptable
6
4
5
120
Acceptable
Standard
7
4
5
140
>125
Standard
7
3
5
125
Acceptable
P equation 2.5 + 0.7 s as suggested by van Herk et al. Target geometry, uncertainties, biological parameters, and TCPpop of references 1 and 2 for prostate are used. Results: O and S become larger when the margin is smaller. O increases (or decreases) one or two units when margins becomes 20% or 40% smaller (or larger) than the RM, similarly S (units increase when clinical tolerance is either CCI-1 or CCI-2). Changes in TCPpop were observed with 20% margin reductions in ref 1. For increased imaging and immobilization efforts, D is decreased. Under standard (RM) conditions, RPN Z 5 x 5 x 5 Z 125. For 20% less than the RM, RPN becomes higher than the threshold, while, CCI-1 with 20% margin reduction will make the RPN less than 125 (6 x 5 x 4). Table 1 shows the RPN’s for other clinical scenarios. Conclusions: An objective method to determine individual margins using risk analysis is developed. The FMEA proved to be a flexible tool to adapt clinical risk in margin determination. Author Disclosure: B. Yi: None. W. D’Souza: None. K.L. Prado: None.
3363 Operations Management in Radiation Oncology: Identifying Workflow Parameters That Portend for Simulation Delay S.R. Blacksburg,1 R. Ghafar,2 S. Green,2 and R. Sheu2; 1Mount Sinai School of Medicine, New York, NY, 2Icahn School of Medicine at Mount Sinai, New York, NY Purpose/Objective(s): Delivering radiation requires the integration of several time-sensitive steps. In some disease states, delay has been found to impact clinical outcome. Despite this, there is a paucity of literature identifying the dependent steps to streamline operational efficiency. This study explores factors that influence time to patient simulation (sim) at a busy academic practice. Materials/Methods: In September 2012, we employed an electronic medical record with automated electronic checklists (ECLs) tied to a rigid workflow. After a sim is ordered, completion of serial steps is recorded; these include insurance authorization, scheduling the sim, and confirming sim appointment. Sim parameters, including desired sim date, body site, technique, insurance, and inpatient status can be cross-referenced. Results: From October 1, 2012 to February 1, 2014, 1534 consecutive sims were ordered; 37.9% requested 3DCRT, 33.5% IMRT/IGRT, and 15.7% SRS/SBRT. 58.2% involved using >10 fractions (fx’s). Insurance providers included commercial (36.1%), Medicare (24.4%), and Medicaid (1.4%). Mean time to insurance authorization, scheduling sim, and confirming sim were 2.2 (0-106), 0.9 (0-76), and 2.0 days (084), respectively. Insurance type (2.4d vs 1.7d for Commercial vs Medicare, p Z .019), >10 fx’s (2.5d vs 1.8d, p Z .01), body site (p<.0001), and inpatient status (1.2d vs 2.3d, p Z .003) influenced time to insurance approval. IMRT/IGRT and SRS/SBRT techniques took longer for insurance authorization than 2D plans and 3DCRT (2.9d, 2.0d, vs 1.2d, and 1.9d, respectively; p Z .003). Mean time to sim scheduling after insurance authorization was 0.9days. Time to confirming sim appointment was influenced by >10 fx’s (2.7 vs 1.1d,