Maximizing Data Assets for Accountable Care

 

A growing number of hospital systems, group practices, and individual health care providers are accepting financial risk for selected populations of patients. As these health systems shift attention to population health management, they are bolstering their leadership ranks with roles such as Chief Population Health Officer (CPHO) and Chief Accountable Care Officer (CACO). With so much at stake, these new executive roles underscore the importance of getting it right and doing it well. This begs the questions, “What challenges are ahead?” AND “What health information technologies are needed to overcome these challenges?” Hint: The answer goes beyond an electronic medical record and a clinical data warehouse.

To profit from value-based payment models, these leaders will need to re-focus people and re-engineer processes to improve coordination of care that yields improved patient outcomes at lower-than-predicted cost. It sounds simple enough, but leading the transformation of both clinical and business operating models will require more than strong political backing and deep pockets; it will require trustworthy and data-driven insights to formulate strategy and make wise decisions about resource allocation, interventional investments, and risk-sharing negotiations. To this end, data-driven intelligence and predictive analytics will offer CPHO/CACOs must-have capabilities.

These leaders will face tremendous challenges to control costs while improving quality. To fully assess and proactively manage the potential risks and rewards associated with value-based payment arrangements, the CPHO/CACO will be highly-dependent on analytics to provide prospective insight and not just rear-view-mirror reporting. Advanced visual and predictive analytics is imperative for assessing individual patients' health risks, identifying gaps in care, preventing treatment errors, improving patient engagement, and controlling costs. For example, by using predictive analytics to identify patients at high risk of readmission, providers can afford to intervene earlier to prevent the readmission and avoid the associated penalties.

To deliver this level of comprehensive, patient-centered care, the care teams in these organizations will need to collaborate more broadly, share patient-level data and begin to incorporate non-traditional data such as psychosocial data, consumer data, genomic data, and digital diagnostic image data. In the not-so-distant future, integrating digital data from audio and video recordings of telemedicine encounters will also supply a wealth of information. The logical extension of this evolution is personalized medicine — holistic, pre-symptomatic treatment focused on preventing costly and potentially avoidable complications. Dr. Mark Selna, Chief Accountable Care Officer at Sutter’s Palo Alto Medical Foundation, said it best, “Ironically, as next-generation statistical modeling tools are applied to ever-growing data sets to infer very nuanced information about smaller and smaller populations, the target population size will become a “population of 1” – which, of course, is the real objective of personalized medicine.”

It would be a mistake to assume that EMRs and other operational platforms could contain all the necessary information, or will evolve from transaction systems into the ‘thinking machines’ necessary to surface analytically-derived insights. Since providers will need to accurately predict patients’ risks in order to direct resources to where they can do the greatest good, integrating patient data from multiple sources is a MUST. It isn't easy, but it IS necessary to deliver the insights that care teams need – to identify the best interventions for each of their patients and to predict the patients’ likelihood of adhering with their care plans and medication regimens.

To realize value from their organization's patient data, the CPHO/CACO will need to overcome the following information management-related challenges:

• Access to and discerning the meaning of comprehensive patient data stored in disparate systems.
  
  
• Access to analytical tools and talent to transform that data into knowledge.
  
  
• Ability to disseminate and take action on analytical insights.
  
  

Each of these challenges will require a robust analytical platform to ingest, store, secure, integrate, transform, visualize, explore, analyze and export patient data. In addition, advanced health analytics requires a rare combination of clinical, technical, and mathematical skills and experience. As such, an extensible platform that leverages the skills of medical informaticists and data scientists to generate and disseminate analytical insights to care teams across any channel or device is both a wise and necessary investment. 

The CPHO/CACO will also need to develop new organizational core competencies to make use of their data and proactively manage risk, such as the ability to:

• Integrate and analyze structured and unstructured data from multiple sources.
  
  
• Interrogate data to continuously reassess clinical risks in near-real time.
  
  
• Understand both clinical and non-clinical factors affecting patients’ clinical risks.
  
  
• Assess the financial impact (gain or loss) of employing resources to reduce these risk factors.
  
  
• Prioritize and deploy resources optimally, for the greatest impact to reduce clinical risks.
  
  
• Recommend prioritized intervention alternatives at the point-of-care.
  
  
• Personalize and automate pre-admission and post-discharge outreach.
  
  
• Inform and facilitate care coordination across multiple channels and facilities.
  
  
• Execute multiple parallel outreach and intervention campaigns concurrently.
  
  
• Measure, monitor, learn, and adapt to optimize overall performance.
  
  

The CPHO/CACO should approach the problem of reducing clinical risks (such as avoidable readmissions, sepsis, and chronic disease exacerbation), by enabling care team members to identify and stratify patients based upon the combination of risk and “impact probability” (i.e. a composite of likeliness to activate to address prioritized gaps in care). By monitoring each individual patients’ clinical risks – in context of their activation level – the care team will be able to understand and manage the key causal drivers, tailor the intervention strategy to each individuals’ specific needs, and execute interventions at the point-of-care and across multiple channels (i.e. text, email, telephone, smart phone apps, etc.).

Progressive health care organizations encourage their patients to become more active participants in treatment decisions and in the management of their care – a strategy often referred to as “patient activation.” Research indicates that patients with the highest level of patient activation have significantly lower costs than those who are the least activated. This approach necessitates greater attention to the role that patients' behaviors play in their treatment compliance. Activating patients—to make informed health care choices and better manage their chronic medical conditions—is a high priority for CPHOs in their strategy to control health care costs.

Having information about their patients’ likely relative activation level can help care teams better manage patients and populations. To accurately score a patient's activation level, though, providers must first gain a comprehensive view of each individual patient from numerous data sources, then assess the patient’s readiness for change, and predict "impact probability." With this information, care team members can personalize care plans for each unique patient and optimize outreach effectiveness by using the best messages via the right channels at the right time.

We know that optimal patient engagement is critical to successful intervention. Therefore, it is necessary to capture and analyze new patient data just-in-time to refresh risks scores and, more importantly, personalize the intervention recommendations. Advanced optimization algorithms should be used to incorporate new information and reprioritize intervention recommendations during patient interactions online, on the phone, and at the point-of-care.

Advanced predictive analytics, visual statistics, and machine learning techniques offer the power to combine analytical insights about patients’ needs with real-time decision support that enables care team members to exercise the most cost-effective and appropriate patient-specific interventions. By using both clinical and socio-economic behavioral data to gain a comprehensive view of each patient, clinicians and care team members can predict propensity to activate, identify preferred communication channels, and tailor the most suitable messages for intervention for each unique individual – at scale.

To be successful, CPHO/CACOs and their organizations must have the ability to take advantage of available information and to overcome information deficits to: prioritize scarce resources for the greatest impact, manage a variety of accountable care risks, and deliver on the promise of value-based care. To accomplish this, they will need a powerful array of data integration, data quality, predictive modeling, optimization, forecasting, and simulation techniques to manage the risks and identify the opportunities to produce the best results.

Avery Earwood is the principal industry consultant at the SAS Center for Health Analytics and Insights