EHR Stakeholder Workshop: Toward New Interaction Models The Nuts and Bolts of Patient Recruitment…from a (nearly) non-technical perspective “What’s right is what’s left when you’ve done everything else wrong.” – Robin Williams “For every 25% increase in complexity, there is a 100% increase in effort” – Scott Woodfield Charles N Mead, MD, MSc Chief Technology Officer National Cancer Institute Washington, DC (USA) Senior Associate Global Health Group Booz Allen Hamilton
1 An Exemplar Scenario… A Trial Sponsor has developed a new intervention for Type I diabetes and has developed a clinical trial protocol to test this new intervention. A repository containing the Electronic Health Records (EHRs) for a number of patients is available to the Sponsor as a possible source of subjects for the protocol. The Trial Sponsor would like to compare the protocol’s inclusion/exclusion (I/E) criteria against patient-specific data in the EHR repository to see how many patients could be potentially eligible to participate in the intervention study.
2 This should be easy except for issues of… Security and Access to EHR repository Consent of individual patient (not necessarily the same as the previous point) Non-standard expression of I/E criteria Non-standard expression of patient-specific data
3 And those were just the ‘easy’ limitations. Also there are… Many additional steps involved in ‘recruiting a subject for a trial’ including More finely granulated analysis of data (beyond I/E criteria) –Lack of standards for automating this analysis, i.e. every recruitment is a one- off process Multitude of regulatory hurdles to cross –Local/State/regional –National/International Multiple stakeholders (with multiple value propositions) working from within multiple systems. For each system involved: –Who mandates a system? –Who pays for a system? –Who uses (primary and secondary) the system? –Who builds the system? –Who regulates the system? Differing levels of organization maturity
4 Complexity “Complicated”, “Multi-faceted”, “Multi-factorial”, “Multi-layered” Ivar Jacobson (paraphrase): “A multi-leveled, vertically hierarchical organization whose products of value are produced through one or more horizontal processes that cross vertical organizational lines.” With cross-organization processes – whether they involve people or systems – syntactic and semantic problems occur at the vertical boundaries. Cumulative experience in industry, art, and (cognitive) science has repeatedly shown that the best way to deal with complexity is through abstraction, layering, and the use of standards.
5 The Communication Pyramid Communication ` Free-text Documents Structured Documents ad hoc Drawings Non-standard Graphics Discussions Standardized Models (UML) Problem Space Solution Space Implementation-Independent Implementation-Specific Abstraction
6 “Protocol” – a ‘commonly used’ term… Source: John Speakman Symbol “Protocol” “We need to sign off on the protocol by Friday” Concept 1 Thing 1 (Document) “Protocol XYZ has enrolled 73 patients” Concept 2 Thing 2 (Study) “Per the protocol, you must be at least 18 to be enrolled” Concept 3 Thing 3 (Plan) Ogden/Richards (Mead/Speakman)
7 A New Interaction Model What is “An Interaction Model”? Candidate definition (CNM): A formal representation of a a set of activities and deliverables that occur as the result of one or more participating entities requesting or responding to well-defined events in a control flow. A given interaction has well-defined –pre- and post-conditions –Inputs and outputs If this sounds like empiric process and/or software engineering, it is… –…but only because software engineering addresses complexity management in situations of equivalent complexity to the proposed goals of this conference Best represented in visual diagrams augmented by text (rather than the inverse)
8 Use Case 2 – Load Lab Data A Formal Representation of an Interaction
9 A New Interaction Model: Critical Components Identify stakeholders by role –Capability, Capacity, Competency –Stakeholders can be systems, organizations, or persons –Many-to-many relationships are common –Five ‘types’ of stakeholders, multiple instances of each type Apply ongoing risk management strategies –Static identification on a regular (e.g. weekly) basis –Integration of risk mitigation strategies into project planning Proceed iteratively and incrementally –Apply project management Best Practices and avoid the Waterfall RUP Agile Scrum Etc.
10 Summary The problem we are trying is the embodiment of a (hyper) complex system apply the appropriate tools, techniques, expertise, etc. –“You can’t build a skyscraper by nailing together doghouses.” The problem will not be solved ‘bottom up’ – a meaningful solution will require top-down mandates to focus bottom-up and middle-out efforts – they will not succeed on their own Success will only occur iterative and incrementally – any attempt to solve this problem with Waterfall approaches is doomed to failure Think architecture: business first, technology second Success in a layered, I/I approach involves –Continuous risk identification and management –Multi-disciplinary teams Identification of discipline-specific value propositions for all stakeholders –Prioritization of project goals and realistic expectation settting The is a hard problem, but it is a solvable one if approached correctly
QUESTIONS & ANSWERS
12 Cumulative experience in industry, art, and (cognitive) science has repeatedly shown that the best way to deal with complexity is iteratively, using abstraction and layering Complex problems require the application of complex cognitive processes in order to achieve meaningful solutions Cognitive processes must apply layering and chunking (“the law of ”) All disciplines that routinely deal with complex problems develop either formal or de facto approaches to Layering and Chunking –Cyclical application of core process of definition, discovery, intervention, (re)evaluation (re- definition) “iterative/incremental process” The Nursing Process as a model of complex problem solving
13 Organizational Maturity Level 1: Heroism and Passion (no defined process) Level 2: A Set of Directions (minimal ability to deal with unexpected) Level 3: A Map (unexpected events can be managed) Level 4: Gathering Process Variance (parallel process improvement) Level 5: Using Process Variance data to drive Process Improvement Everyone wants to be Level 5 Progression to the ‘next level’ is stepwise Level 1 does not mean incompetence! It just doesn’t scale well over time
14 Complexity “Complicated”, “Multi-faceted”, “Multi-factorial”, “Multi-layered” Ivar Jacobson (paraphrase): “A multi-leveled, vertically hierarchical organization whose products of value are produced through one or more horizontal processes that cross vertical organizational lines.”