C ANCER C ARE E NGINEERING The Health System - Systems Engineering - Health Services Research - Medical Informatics - TRIP - Policy - Economics INDIVIDUAL PATIENTS - Quality Care - Epidemiology - Decision Support - Detection - Symptoms / QoL - Access Biological Sciences - Blood Biomarkers - Tumor Biomarkers/Dynamics - Clinical Trials - Susceptibility - Environmental Factors - Tumor Variations mutations sessile polyps Accelerated translation of science to practice Screening Rates Identified best practices to dramatically improve and personalize tx Better decision tools that aid providers, managers A knowledge management system incorporating latest research advances so that every piece of new knowledge does not have to be manually assimilated by every provider Methods of implementation and systems engineering to address systems complexity and speed of response 01001 00100 01001 00100 01001 10100 11010 01011 01101 Matt Burton, MD Knowledge of Best and Actual Practices Clinical Pathways, Guidelines, Order Sets, Quality Indicators Clinical Workflow Prediction, CDSS, Order Management, Communication, Care Delivery, and Documentation Simulate, Monitor, and Analyze Clinical Data Results Delivery, Quality Measurement and Monitoring, Feedback/forward (CDSS, Simulation), Knowledge Discovery Virtual Hospital and Cancer Care Engineering Matt Burton, MD Focus on Translation Goal-Oriented Drive improvements by using key metrics that summarize system behavior, such as the NIH statistics cited above; Global Awareness Systematic collection, analysis, and dissemination of cancer system data to all participants for purposes of more effective distributed actions by system participants; Metric Oriented Direct researchers towards overcoming barriers likely to result in the greatest care system improvements; Knowledge Improvement Distill research knowledge to useful products that can be easily used by providers, consumers, and others to overcome the greatest complexities in cancer care Understand and direct work by considering it within a system wide perspective; Externalize Knowledge Reduce knowledge to models that can be widely learned and whose properties can be tested for improvement. System Level: Indiana Regional Cancer Care System Simulate the Indiana colorectal cancer (CRC) care system to resemble current performance Using the model: Identify areas of potential improvement in current system Develop and test various strategies to arrive at an optimal strategy 6/17/2008 cceHUB Investigators: Selen Aydogan-Cremaschi, PhD, Purdue Discovery Park Brad Doebbeling, MD, MSc; Seza Orcun, PhD, Purdue Discovery Park, David Haggstrom, MD, MAS, Multiple others S. Cremaschi 7 Interview providers to identify and rank questions of interest 2. Develop CRC care system models and implement them 3. Validate the model 4. Run what-if scenarios to answer questions of interest 1. 6/18/2008 cceHUB 8 9 Survivor Population disease free Cancer Population General Population negative cancer or care complication or other not treatable or do not wish to be treated cancer or care complication or other Death cancer or care complication or other Screening Symptomatic cancer or care complication or other Treatment Staging & Evaluation positive Diagnosis false positive 6/18/2008 cceHUB 9 1. Identify higher impact components of the CRC care system in Indiana: a) From a healthcare system perspective, what indices or metrics might be early warning signals for managers or clinical leaders of where to intervene quickly? b) If we consider the entire spectrum of care, can we have the greatest impact on CRC care mortality and cost of care by optimizing one of the components to perform in a highly reliable fashion? : --Screening, Screening Follow-up, Diagnosis, Treatment - early stage, late stage diagnosis, survivorship, palliative care 2. Determine necessary system resource capacities: a) If every positive abnormal screening test is followed up with a colonoscopy, does Indiana have the necessary resources? b) What should the capacity of the high-volume facilities be in order to be able to perform the necessary surgical procedures for CRC? 6/17/2008 cceHUB S. Cremaschi 10 Natural Development and Progression of CRC in Population Agent Prob. of Dev. Polyp Normal Invisible Polyp Distant CRC Regional CRC LOP Dist. between States 6/18/2008 Polyp < 1cm Polyp > 1cm Local CRC Symptomatic CRC cceHUB In Situ CRC LOP Dist. from asymp. to symp. 11 Screening & Follow-up Screening Choice Compliance Intervals FOBT Every 1 year Sigmoidoscopy Every 2 years Colonoscopy Every 5 years Never Compliant One Time Compliant 6/18/2008 cceHUB 12 Treatment Type Number Combinations Adherence Date Result Lifestyle 6/18/2008 cceHUB 13 Attributes Years in training Provider Type, Specialty Location/County Volume of patients/procedures Adherence with “Evidence-Based Medicine” Treatment maps for CRC stages using the interviews of CRC providers/specialists. 6/18/2008 cceHUB 14 Attributes Location Resources ▪ Screening/Diagnosis/Staging ▪ Surgery ▪ Radiotherapy ▪ Chemotherapy ▪ Hospice ▪ Palliative Volume of patients/procedures 6/18/2008 cceHUB 15 System Level: Indianapolis Clinic System Investigators: PI – Brad Doebbeling, MD, MSc; Co-PI –Jamie Workman-Germann, Purdue University School of Engineering and Technology at IUPUI Anticipated Outcome: (1) Develop cancer prevention and care process maps and quality reports for Indianapolis clinics, (2) understand barriers to best practice care, (3) utilize a Cancer Care – Technical Assistance Program (TAP) to help implement best practices in one or more clinics. CRC Systems Redesign Project Aims: ▪ To understand EMR implementation impact of clinical processes ▪ To optimize patient flow by identifying process barriers to screening ▪ To increase CRC screening rates by removing clinical barriers Methods: ▪ Facilitation of interdisciplinary teams of IUMG and VAMC primary care clinic staff and area supervisors ▪ Evaluation of existing clinical workflow ▪ Systems engineering, Lean, Positive deviance principles Scope of Work: Assessment of Primary Care CRC processes Development of data infrastructure to support sustainability of initiative Develop and administer cultural assessment to determine organizational readiness for IT and systems redesign initiative implementation. Facilitate IUMG and VAMC project teams in application of systems engineering tools to: --Design and perform a pilot test of the process redesign, ensuring new processes meet clinical and economic objectives, timeline requirements, and project deliverables --implement new processes and systems with a robust control strategy to ensure long term sustainability of improvements System Level: Cross System Information Awareness and Clinic Investigators: PI - Caroline Carney Doebbeling, MD, MSc, Associate Professor of Medicine & Psychiatry, IUSM; Research Scientist, IU Center for Health Services & Outcomes Research, Regenstrief Institute, Inc.; Director of Quality and Outcomes, Indiana Medicaid Co-PI – Katherine Schilling, MLS, EdD, AHIP, IU School of Library and Information Sciences and IU School of Informatics Specific Aims: (1) Develop a best practice CRC treatment map including recognition and treatment of cancer-related distress; (2) identify opportunities for streamlining processes; and (3) pilot implementation of more efficient delivery of psychosocial services (mental health and social work) concurrent with cancer treatment. Long-term objective is to implement innovative patient screening and navigation systems to consistently deliver optimal psychosocial care. CCE-3 – Literature Matrix Universal screening in all clinics Distress Thermometer Triage to appropriate CL providers - Moderate to Severe Distress: Mental health, social work, pastoral services - Mild Distress: Primary Oncology team Barriers - Too few providers - Waiting lists Recommendation : Can we identify, describe, and better understand “positive deviant” systems within treatment centers nationally that are engaged in best practices? How have they been successful in implementation? CCE-HSR C. Carney Doebbeling System Level: System Biology-Oncologist-Patient Investigators: PI – Seza Orcun, PhD, Purdue Discovery Park Co-PI – Doraiswami Ramkrishna, PhD, Harry Creighton Peffer Distinguished Professor of Chemical Engineering, Purdue Co-I – Eric Sherer, PhD, Purdue Discovery Park; VA Center of Excellence on Implementing Evidence-based Practice; Tom Imperiale, MD, Professor of Medicine, IU School of Medicine and IU Center for Health Services & Outcomes Research, Regenstrief Institute Anticipated Outcome: (1) Develop a population balanced model to predict efficacy of oncology treatment, (2) validate model with oncologist usage, and (3) engineering modeling researchers in clinical settings partnering on joint projects with oncologists, GI specialists and services researchers. 1. CRC prevalence model that includes intermediate polyp states and tumor genetic heterogeneity 1. Already several similar models for incidence 2. None (that we know of) that include polyps or branching 2. Methodology to extract a minimal set of discrete patient model parameter sets from CRC & polyp prevalence / incidence data 1. Parameter sets are independent of demographics => Bayesian model for predicting likely parameter sets for an individual patient 1. Certain demographics may be more likely for certain parameter sets 2. Likelihoods adjust to additional patient information 3. Predict incidence, treatment outcome, outcome 05/28/08 CCE-HSR E. Sherer System Level: Cross System Information Awareness – Assimilation and Integration of Data From All Projects Investigators: PI – David S. Ebert, PhD, Professor of Electrical and Computer Engineering, Director of Purdue University Regional Visualization and Analytics Center, Director of Purdue University Rendering and Perceptualization Lab, Purdue Primary Objective: Full-fledged, interactive, integrated visual and statistical analysis capability in a vital analytic environment that brings together massive, disparate, incomplete and time-evolving -omic data sets. Longer term goal---linkage with systems level data—cross projects with EMR, claims, structure, process, outcomes “The Dashboard Project” Initial pilot, creation of an interactive, integrated dashboard of facility-level colorectal cancer performance measures to inform the process of cancer care and systems management in the VAMC. An example of functionality would be the ability to view CRC-related, facility-wide data output by clinics, treatment providers or risk-level of patient populations. Brad Doebbeling, MD, MSc; Selen Aydogan-Cremaschi, PhD,; Matt Burton, MD; Timothy Carney, MPH, MB,; Jason Saleem, PhD; Darrell Baker, RN; David Haggstrom, MD; Tom Imperiale, MD; Charles Kahi, MD, and Chris Suelzer, MD Dashboard Report – Corporate View 06/08 Critical, Clinically-Relevant Questions of Interest: What percentage of patients who have received physician-ordered FOBT cards, are not returning them? What factors are contributing to noncompliance? Regarding follow up after a positive FOBT screen, what percentage of patients is notified within the required 14 days of the results? What percentage of patients with colonoscopy orders to follow-up for positive screens isn’t getting colonoscopy completed? What factors are contributing to this gap? What percentage of patients with a positive screen get needed colonoscopy within the required 30 days? CCE-HSR B. Doebbeling -Questions of Interest -Project kick-off -Interview VAMC CRC care providers & administrators -Identify critical questions of interest -Identify CRC performance measures for visualization -Conceptual Design of Dashboard -Prototype Implementation and Testing -Mock Implementation Feb 2008 Apr 2008 -Review -Literature for dashboards for healthcare system -Implemented healthcare dashboards May 2008 Aug 2008 June 2008 -Detailed Definitions of CRC Performance Measures INITIATION PHASE Oct 2008 Sep 2008 June 2009 Dec 2008 -Database & Software Selection -Usability Testing & Dashboard Refinement PHASE I -Dissemination -Proposal Development PHASE II System Level: Repository of Data for All CCE Projects, Preparation of Data From All Projects, and Strategic Statistical Analysis Investigators: PI – Marietta L. Harrison, PhD, Purdue University; Co-I, Laura Jones Myers, PhD, George Allen, IU School of Medicine & VA COE Anticipated Outcome: (1) Utilitarian project to provide an electronic repository of all CCE project data, (2) Develop a procedure and tools for cleaning and validating CCE project data, and (3) Determine a strategy for combining and analyzing the disparate data from all projects Organizational Social Subsystem Structure / Work System Design Joint Optimization External Technological Subsystem Environment Aim 1: Identify key approaches to CDS development for CRC screening at two VAMC sites and two nationally recognized nonVA sites, for effective CDS integration into clinical workflow. Aim 2: Develop and test CDS design alternatives for improved integration into clinical workflow through a controlled simulation study and subsequent implementation. Research Team: Brad Doebbeling, MD, MSc (PI); David Haggstrom, MD, MAS; Jason Saleem, PhD ; Laura Militello, MA; Heather Hagg, MS; Shawn Hoke and Lori Losee, and West Haven VA, Columbia, South Carolina VA, 30 Partners Healthcare (Harvard). Training and research components: Aim 1: To identify patient-level characteristics associated with underuse & overuse of surveillance care among CRC survivors colonoscopy, CT scans, CEA tests, history & physical Aim 2: To determine whether organizational or physician characteristics are associated with the quality of CRC surveillance care Aim 3: To develop and test a CRC survivor’s personal health record that promotes high-quality follow-up care Develop methods and tools for effective use of unstructured data such as narrative text in VA EHR Improve text processing, text mining and de-identification capabilities Applied projects Seven VAMCs with informatics research capabilities participating--Salt Lake City, Nashville TVHS, Indianapolis, Palo Alto, Portland, Tampa, West Haven. Also Boston – MAVERIC, Pittsburgh-Philadelphia, Mayo Clinic, Carnegie Mellon. Ability to extract free (tumor stage, etc) through a web services model. [See Len D’Avolio, Mahesh Merchant, Matt Burton] Innovation, Interdisciplinary Collaboration, Focused on transforming Healthcare Positive impact on healthcare Reduction in cost of healthcare delivery Increase in the value of healthcare delivery Translating the project results to the benefit of the healthcare system Possibility to leverage the Foundation, DoD, VA and AHRQ funding Knowledge Patient with given demographic needs a Colonoscopy every 10 yrs Clinical Workflow Pt J. Doe scheduled for routine H&P plan for Colonoscopy in next 3 mos. w/ PCo Clinical Reminder: “Dr. Smith, J. Doe needs a Colonoscopy” CPOE: “Colonoscopy for J. Doe is ordered and signed” Order Mgmt: “Colonoscopy for J. Doe scheduled on 7/12/08” Resource/ Supply Mgmt: “Need Colonoscopy suite, resources, and supplies on 07/12/08 at 9:00AM for Pt with give requirements” Registration: “J. Doe has arrived for his Colonoscopy” Document Preparation: “Populate fields in Procedure Note for Colonoscopy” Documentation: “Colonoscopy begun/ completed at 9:03 AM/ 9:37 AM 7/12/08” Order: “Path Specimen for Polyp” Document Preparation: “Populate fields in Path Report on Joe Doe’s Polyp” Clinical Data Preliminary Procedure Note for Colonoscopy on 7/12/08 is resulted Pathology Report on Colonoscopy on 7/12/08 is signed and resulted Matt Burton, MD System Level: Indiana Regional Cancer Care System Investigators: PI – Selen Aydogan-Cremaschi, PhD, Assistant Research Scientist, Purdue Discovery Park Co-PIs – Bradley N. Doebbeling, MD, MSc; Seza Orcun, PhD, Associate Research Scientist, Purdue Discovery Park Anticipated Outcome: A model that can explain the existing CRC care system data, answer “what-if” questions about potential changes to the care system, and suggest improvements based on analyzing various options. Mechanistic modeling of colorectal cancer (CRC) Includes genetic mutations and growth / death dynamics Hypothesize mechanisms ▪ Underlying knowledge ▪ Level of detail determined by measurements ▪ Can be extended to incorporate additional information as it becomes available 05/28/08 Prediction of likely individual patient CRC Temporal likelihoods of CRC Likely properties of CRC CCE-HSR E. Sherer System Level: Clinical Project Team: PI – Brad Doebbeling, MD, MSc Co-Is – Selen Aydogan-Cremaschi, PhD, Assistant Research Scientist, Purdue Discovery Park, VA HSR&D COE; Matt Burton, MD, Medical Informatics Fellow, Regenstrief Institute, Inc.; Timothy Carney, MPH, MBA, IU School of Informatics; Jason Saleem, PhD, Assistant Professor, VA COE and IUPUI School Engineering & Tech; David Haggstrom, MD, MAS, Assistant Professor, IU School of Medicine, VA CIEBP and IU CHSOR Consultants – Darrell Baker, RN, Clinical Applications Coordinator, VAMC;; Tom Imperiale, MD, Research Scientist, Regenstrief Institute, Inc. and IU School of Medicine; Charles Kahi, MD, Roudebush VA Medical Center and Chris Suelzer, MD, Associate Chief of Staff for Ambulatory Care, Roudebush VA Medical Center More effective use of IT is recommended in integrating point of care access to (e.g., Committee on Quality Health Care in America): Health literature and evidence-based guidelines; Computerized clinical data; Computerized decision support (CDS) systems; Automation of decisions to reduce errors; Electronic communication among providers and patients into practice. Computerized CDS can improve clinician decision making and support adherence to evidence-based guidelines. Colorectal cancer screening focus: high disease burden, relatively low screening rates, strong evidence for screening effectiveness Failure to optimally integrate CDS into workflow has resulted in inconsistent and incomplete implementation strategies. Disparities in hospital selection Cultural disparities Gender & Age Diagnosis at younger age = higher risk for psychosocial 05/28/08 problems related to illness burden Racial disparities (??) Literacy and Health Literacy Low health literacy associated with less knowledge about colorectal cancer Low health literacy associated with less knowledge about screening Practical, life-management issues (insurance, employment) Lower income, Medicare Part D issues Ability of providers to recognize distress CCE-HSR K. Schilling System Level: Physical CRC Sample and Raw Data Collection Investigators: PI – Stephen D. Williams, MD; Gabi Chiorean, MD, IU Simon Cancer Center, IU School of Medicine Anticipated Outcome: Augments a DoD-funded project to collect additional samples and clinical data based on an analysis of initial results. To assess physical samples and laboratory analysis data from the CPTAC project and other Indiana based cancer specimens for purposes of understanding Indiana’s capacity for a total cancer care engineering project. System Level: Opportunistic Project Refinement and Integration of Regional and National Assets into Indiana CRC CCE Effort Investigators: PIs – Joe Pekny, PhD, Purdue University; Brad Doebbeling, MD, MSc, Indiana University Primary Objective: Management of the portfolio of all projects to maximize impact and to leverage success by further incremental investment