As the U.S. healthcare industry as a whole moves forward towards fundamental change, replacing a volume-driven, fee-for-service-payment-based system with one focused on accountable care, population health, transparency, and value-based payment, both health insurers and providers are finding themselves compelled forward towards new definitions and new horizons. It was in that conext that HCI Editor-in-Chief Mark Hagland interviewed Charles Kennedy, M.D., along with other healthcare leaders, for the May/June cover story of Healthcare Informatics.
Kennedy is the chief population officer at Healthagen LLC, a subsidiary of the Hartford, Conn.-based Aetna, and a health insurer executive helping to guide dozens of accountable care organizations (ACOs). Aetna has been moving forward assertively to collaborate with hospitals and physicians to develp private-market accountable care organizations (ACOs), which generally match the ACO concept evolving forward within several federal programs under the aegis of the Medicare program (the Medicare Shared Savings Program, Pioneer ACO Program, and now also the Next Generation ACO Program).
Hagland interviewed Kennedy for the May/June cover story. Below are excerpts from the extended interview conducted earlier this spring.
What are the greatest challenges involved in trying to leverage data and data analytics for population health work?
The overarching goal of accountable care and value-based payment programs is to move the national payment systems toward rewarding the value of care delivered rather than the volume or complexity of care. The ideal value-based care framework does two things. First, it allows providers organized in an ACO or similar structure to take accountability for each patient in a population. And second, it encompasses the tools to help providers and their patients receive the right care at the right time and place. Done right, these models lead to better health of the individual and the population.
One of the most significant challenges to moving to value-based care is complementing the current approach to quality measurement, which is largely retrospective and based on post hoc statistics about the aggregate cohort of patients. Many dashboards, quality measures and electronic records today provide retrospective views and value. What we need now is the real-time insights on a patient’s health, including where they’re going for services and how compliant they are with their care plans, so that we can act on information at the time it will most make a difference in the patient’s care.
And it is challenging to leverage currently available information systems for these purposes, correct?
Yes. Health IT systems that offer the equivalent of a clinical navigation system are woefully under-deployed. Achieving the required information state requires a new HIT infrastructure, supporting integrated administrative, claims and clinical data from all sources reorganized and optimized to assist with value-based care interventions for each individual’s health and care. These records must be not only semantically interoperable, but must also be structured in such a way that provides useful and usable information on each individual patient. Today, however, many EMRs function like electronic file cabinets.
The kind of understanding and usefulness I describe requires advanced knowledge engineering such as ontologies that are purpose-built to re-structure data collected by providers and payers, leverage such knowledge to allow rapid (measured in hours to days not weeks to months) connectivity of any system with information about patients, and provide the ability to trigger actions to assist patients and providers at the time such actions matter. Taking data from lots of sources and lumping it under a patient so that each provider can try to figure out what is going on with the patient each time they need to make a decision is not semantic interoperability.
That really requires moving process closer to being real-time and point-of-care whenever possible, correct?
Yes. Trying to fix all of this after the fact by hiring legions of nurses as care/case managers to try to manually assemble such records and then act upon them using phone calls to doctors and patients and then to use their own clinical skills to attempt to improve the care of the sickest 5 percent of patients not only ignores the other 95% (who are inexorably moving toward the top 5 percent if they are chronically ill), but is too expensive.
Value-based care requires the ability to synthesize all of this information into an easy to consume record of the patient that allows a caregiver, care manager, doctor, or patient to take an action differently than they otherwise would and results in higher patient value. New IT frameworks now exist that perform this function and are just starting to be deployed by health delivery organizations and health plans. The lack of this kind of infrastructure in most organizations still represents the greatest challenge in moving to value-based care.
What are the greatest opportunities in value-based care?
The greatest opportunity in value-based care is the ability to support time-of-care health delivery, simultaneously involving physicians, other providers, patients and their casual caregivers. This will become possible through patient-centric, ontologically-based, real-time infrastructures.
Today what we have are EMRs and HITs that really are merely electronic documentation tools. The data is not well structured (if it is structured at all) and so providers spend tremendous amounts of time entering data, but get little value back from all of that effort. The latest Rand/AMA report underscores this need in a number of ways.
Providing the right IT support to refocus the health system will make care more efficient and effective than it is today – while improving the satisfaction of providers and patients who are far more interested in improving health than in dealing with yet another mountain of documentation artifacts.
What have been the biggest lessons learned so far in the use of analytics and data, and collaboration
in this area?
In addition to the need for a new value-based infrastructure to support the entire effort discussed above, our biggest learnings have been: understanding and dealing with the limits of today’s data sets; properly emphasizing the importance of process re-engineering; careful program integration, and the challenge of finding appropriate staff.
Claims have information not available anywhere else, but if claim data is introduced into clinical settings, it must be intelligently harmonized with the medical record information received from clinical systems – generally considered by providers to be the gold standard. If it is not, it will be rejected by the providers.
Having effective staff that understands how to leverage clinical data and claim data, understands the strengths and weaknesses of each data set, and then can put the resulting insights to work for value creation (at multiple IT and clinical levels) is not a trivial skill. Further, these individuals need work processes tightly embedded in care delivery and the supporting IT infrastructure. Dealing with any of this as a “tack on” generally just creates more work for all without much realized value.
Finally, to collaborate with providers effectively, trust is the foundational challenge. Health plans starts with a bit of a disadvantage here, in part due to the historical win-lose contracting dynamic. However, trust can be achieved through transparency, shared goals, the joint use of tools that providers and payers can rely on, and incessant attention to alignment.
Where is the industry in terms of maturity of tools and infrastructure?
In my experience, most of the organizations doing “population health” in clinical settings have created one-off solutions, some of which work in their unique environment. Nearly any attempt to improve care can find low-hanging fruit that yields cost savings and/or quality improvements – even entirely manual systems of improvement. However, almost all of these tools prove difficult -- if not impossible -- to translate/transfer to other clinical settings.
Historically, health plans and disease management companies have focused on population health from a payer perspective – generally deploying banks of nurses to outreach to their members that have specific conditions or risk factors. The opportunity for health plans is to take this infrastructure and re-direct it by integrating it with providers and delivery systems. The increased adoption of new infrastructure tools designed to integrate the clinical and financial environments around each patient will provide the grounding to rapidly mature value-based care. Coupled with increased buy-in by the clinicians involved in providing such care, this cannot happen quickly enough but if and when it does, I think significant value for all involved will occur. And the patient will be the one who benefits the most.
In the end, where are we as an industry right now with this, on a big-picture level?
The bottom line is that we are in the infancy of value-based care, often still conflating improving individual care and thus improving a population with gathering statistics about a population and wondering how to use those to improve care of individuals. As long as there is no fundamental understanding of the information and ability to integrate such understanding easily across all sourcing systems at the time of care, the maturation of such approaches will be slow. But there are new infrastructures now emerging that enable these functions. I am cautiously optimistic that these next generation tools coming to market will finally provide the resources to deliver on the promise of health IT and help our nation realize the vision of high value health care.