OVERVIEW ACLC s Mission: Accelerate the adoption of a range of accountable care delivery models throughout the country ACLC s Vision: Create a comprehensive list of competencies that a risk bearing entity will need to succeed in a valuebased payment environment ACLC Patient Risk Assessment and Stratification Workgroup In-Person Meeting Objectives Review domain name: Decide whether the name is appropriate or should be revised. Identify and Define Competencies: Specific competencies for each category will be refined. Terminology: Terminology will be identified, defined, and refined as needed.
PROGRESS: The progress related to each objective is outlined below. Domain Name The in-person participants reviewed the domain name and considered the meaning of both risk assessment and stratification. Risk stratified care management and care coordination are essential components to value based accountable care. In this context the word risk has been loosely used to mean the risk that the patient will require additional, costly and sometimes unnecessary care. All too often, this means only the risk that that the provider or the organization will be financially disadvantaged by penalties for readmissions or costly services that are preventable and drive up the total cost of care. Because of this emphasis on cost, many of the early models for stratification have focused on analysis of claims data with little if any clinical input. More recently we are using all available data but the emphasis remains on cost control. The work group would like to redirect this conversation to be more patient centered and align with the task of providing specialized care and services to individual patients. With that in mind we would recommend that the process be named: Patient Assessment and Determination of Need. Domain Definition/Description The essence of Patient Assessment and Determination of Need is to identify if an individual needs help managing his or her chronic condition (care management) or needs help navigating the complex and often fragmented system with multiple points of care (care coordination) or both. Assigning a patient to an appropriate category of need allows for the use of specific strategies and resources allocation to better serve and provide value at the patient level. The desired outcome is to identify interventions that are most likely to improve patient care and patient oriented outcomes spanning a wide variety of potential interventions. The model should inform the choice and level of intervention. The development of patient assessment models should parallel and align with the activities that will eventually constitute care management and care coordination rather than be integrated afterward.
Categories Four categories were initially proposed to the workgroup and the same four essential categories have been recommended as the essential categories in Patient Assessment and Determination of Need. The table below identifies each proposed category, category description, and the essential competencies. Categories, descriptions, and competencies will continue to evolve as the larger workgroup is able to review the work to date. Tentative Proposed Categories Platform for Patient Assessment Category Description The platform is the system of hardware, software and processes which enable and support analysis, visualization and best practices to assess and stratify a population with the specific intent to identify specific care needs and/or opportunity for health outcomes improvement, at both the population and individual patient level. Essential Competencies (representative, not comprehensive) Support the use of multiple common analytic tools via an open API Support multiple user/organization defined stratification and risk as well as proprietary and open source models Support multiple data types including administrative, clinical and patient reported data in a single comprehensive data warehouse for effective analytics Support multiple levels of analysis from the entire population to an individual patient Enable analysis of targetable behaviors and interventions based on specific needs of multiple program models
Enable user defined variable weights and models for multiple care models or programs Support a wide diversity of the population Support multiple risk assessment models based on business use or need Comments: Data Stand-alone disease specific interventions (e.g. a diabetes registry) can inform the systematic care of a sub-population of patients with diabetes. Most patients have multiple chronic conditions which requires information about all of the conditions at the point of care (e.g. a patient centered registry) Patient-specific sociodemographic, clinical and behavioral health information collected from a variety of sources that serves as the basis for determining a patient s state of health and the level of financial and medical risk associated with that state. Evaluate and use demographic information Review and use claims and utilization data Review and use clinical data, including lab and pharmacy Encounter data (not same as claims or utilization data) Additional competencies (beyond essential) Level 2: Important, but not essential Evaluate end of life status Evaluate behavioral/mental health history and status Use patient's functional status in stratification Review patient's potential physical limitations Socioeconomic Information Level 3: Nice to Have Incorporate referrals from a variety of outside sources, including care
Competency Additions: management, community resources, hospital discharge planner, family, etc. Incorporate member-reported assessments and information Provide and evaluate real time information Clinical Notes Data from mobile devices Review and use physician input Use publicly available data sets for geographic high-cost concentrations Identify diagnoses that drive spending Use the Patient Activation Measure as a predictor of risk Use population based tools Refresh information frequently Incorporation of patient centered assessment of health state Comments: Implementation and Data Processing Encounter data competency need more wording around it not clear what it is Implementation and Data Processing is the real Evaluate the data input for the algorithm time run of live data through the platform in Combine analytics with intuition, and look order to determine the patient risk index based beyond high utilization when stratifying in order on predefined rules and algorithms in order to to address the patients needs propose a personalized best care pathway for Develop a flexible system to adapt to any comorbidity model & Leverage GIS mapping in the quality treatment of each patient. assessment and stratification
Ensure appropriate staffing to run data and reports Test the data periodically to assess tool accuracy Integrate data with evidence-based guidelines Develop automated analytics for Dashboard Create a dashboard for health condition prevalence and practice performance Identify population segments with modifiable costs, and monitor risk for improvements in care pathway Determine targetable behaviors and interventions based on program model Provide and evaluate real time information and refresh frequently Competency Additions: Explicit strategies commensurate with the level of patient need to ensure optimal outcomes. Comments: It seems like this is the place to emphasize to conversion of data to useful knowledge and intelligence to inform optimal patient care. May need a sexier title.:-) Proposed definition change: Implementation and Data Processing is the real time run of live data through the platform in order to determine the patient risk index based on predefined rules and algorithms in order to propose (or inform care management and care coordination) a personalized best care pathway for the quality treatment of each patient. Re: proposed definition above: Ultimately none of these domains operates in a vacuum, but risk and need evaluation does it prescribe the best pathway or does it feed into those domains
Patient and Population Risk Assessment Dashboard that specifically address care (both management and coordination)? Perhaps I m splitting hairs here. OR, perhaps this goes below in Cat #4 A meaningful and actionable report providing Should identify members prospectively (using both population and patient level data to available retrospective data) with varying levels prioritize interventions by levels of risk and of risk identify areas of needs that can improve patient Should be able to be segmented by filters (i.e. care and patient-oriented outcomes. The payer, provider, health condition, psychosocial dashboard will inform the choice of or behavioral health, etc.) intervention(s). The dashboard will not only Should be user-friendly, interactive and easily provide risk stratification but also provide modified to meet the needs and resources of the organization patient-level drivers for each level of risk. Should be flexible enough to allow for reevaluation of risk/needs based on Care Manager/Care Coordinator input Comments: Category name: We may want to drop out the words risk assessment here and call it just Patient and Population Dashboard Category Definition: The term dashboard in common parlance refers to a visual display of a set of metrics over time (i.e. Run charts) presented on a computer display or ideally as a series of printed graphs hung on the wall for all to see. I think this concept should be incorporated into the definition. It is not intended to be just a report that may or may not be viewed by all. Maybe adding one or two competencies would deal with this concern.
Terminology Potential Term for Review Descriptive Analytics High Risk High Needs Population segmentation Predictive analytics Predictive Modeling Possible Definition from Literature Review Descriptive Analytics looks at data over time to identify trends, spot gaps in population health and improve patient outcomes. Predictive Analytics leverages modeling tools to identify the subpopulations most at risk. "Predictive models are designed to identify patients of the highest risk in Definition Proposals Vs. inferential analytics?? High risk for providers-the risk that a patient is likely to incur avoidable costs. High risk patients- The risk that the patient needs additional help managing his or her chronic condition or navigating the multiple points of care. a person with a high risk for health care services Patients who need additional help managing chronic conditions (care management) or navigating a complex system with multiple points of care (care coordination) a person with a high number of current health care needs Segmentation of the patient panel for the purpose of applying specific strategies or interventions to prevent avoidable complications or unnecessary utilization. a way of bucketing or stratifying a population of people according some set of criteria, such as: risk, needs, geography, medical condition, age, etc. Some overlap here with prospective risk modeling, prospective algorithms and descriptive analytics. Add the word proactively to definition on left a tool which uses regression analysis or some other mathematics based method to predict the occurrence of some future event or condition This sounds the same as risk stratification Add the word proactively to definition on left
Prospective risk modeling Prospective algorithms Risk Assessment Risk stratification Patient Segmentation Population health data analytics a population and to provide information useful in supporting care and health management. Risk assessment is the measurement of that risk, linking the characteristics of an individual to their current and future resource use or clinical outcome. a mathematical model, often based on regression analysis which can be used to predict the occurrence of some future event or condition Using all available data to proactively address the potential for deteriorating health status Using predictive analytics and modeling with respect to the risk or need for future health services at the person or population level This may be synonymous with Prospective risk modeling above This current definition is from the frame of the provider organization that is trying to avoid unnecessary costs or complications. A patient centered view would be more aligned with "high needs" above. Use this terminology to define High Risk / High Needs above. The use of information known about a particular person to determine their risk for future health care services A systematic method to identify individuals who need help managing their chronic illness (care management) or help managing their care plan with multiple points of care (care coordination) Using a risk model (defined above) to bucket individual persons from a population together by risk level This may be synonymous with Population segmentation above. Assessing the result of population health strategies using claims data, clinical information and patient oriented outcomes of care. Analytics regarding the demographic, socio-economic and clinical health of a population of persons
Stratification allows direction of program resources most appropriately The creation of various levels of need within a patient panel so that resources can be appropriately allocated in proportion to the patient specific need. Additional Terminology Proposed Data Lakes Visualization Dashboard Utilization Population Patient Stratification A data modeling method which captures large amounts of data without the need to create a specific data model at the time of data acquisition, often associated with "big data" The layer of software or application that is used to present data via tables and graphs to an end user A high or top level view of analytics data and metrics across an entity or organization, may or may not be supported by additional "drill down" capability to more granular information Use of healthcare services by a patient, group of patients or population A group of person for which an organization has some level of responsibility in terms of health outcomes including quality and cost. Some number of persons will require health care services these are patients, other persons will not experience the health care system but would be eligible for health and wellness programs or activities The use of demographics, clinical data, socio-economic data and one or more risk models to group similar patients.