Predictive

An ounce of prevention is worth a pound of cure. The earlier we can identify a person's needs and appropriately match those needs with available and cost-effective resources, the better the outcomes. Care avoidance due to cost, inconvenience, fragmentation, confusion, access barriers, and competing priorities can lead to delays in diagnosis, poor health, lower quality of life, and downstream use of the most expensive and least fun parts of the health care system, like the ER and the hospital. 

I'm a hospitalist, so I know that people generally don't want to see me. I often remind myself that when I walk past those sliding glass doors, the people I'm caring for are having some of the worst days of their lives. Perhaps the clearest example of investing in upstream management to prevent members from needing to meet me in the hospital is advanced primary care. Findings from a JAMA Open Network study between Collective Health and One Medical suggest that individuals who used comprehensive primary care services had lower total costs of care within a plan year, despite higher up-front primary care spending. And yet, despite a clear value proposition for both better health and lower costs, primary care remains vastly underused, with a quarter of Americans lacking a routine source of care. 

How do we reach those who are not engaged and get them plugged in to the right parts of the system? Predictive engagement aims to match outreach and resource allocation to member needs as intelligently as possible. The idea is that getting the right information to the right person at the right time will help patients navigate our fragmented health care landscape in a more cost-effective manner because it is personalized and actionable. The importance of timing is key here. 

Being proactive and getting in front of issues is crucial to successfully scaling personalization and reducing wasteful spending. Implementing new technologies, such as artificial intelligence, machine learning, and predictive analytics — and ensuring they are applied equitably and securely — are central to the scaling strategy. The patient's information, properly filtered through predictive analytics software, can become one of the most powerful tools in effective care management.

Precise 

To truly harness the power of predictive and personalized analytics, our care suggestions must be precise. In my view, the importance of precision extends from the vital nature of trust in improving health. We need to get it right at every touchpoint, because getting it wrong erodes the trust we need to truly serve member needs. The current health care system is riddled with errors and dropped balls, from claims processing and billing, to diagnosis and treatment, to scheduling and follow-up. We tend to focus on ‌missed or delayed diagnoses, but casting too wide a net can also be problematic. Imagine the stress of receiving a communication suggesting that you have cancer when in fact you do not. Good luck seeing "the big C" show up on your phone and then getting back to your workday. 

Although these recommendations are a powerful tool, we can take them a step further with a prioritization scheme for each member's list of recommendations. As a clinician, I think about specific patient examples, like a person with diabetes complicated by peripheral vascular, heart, and kidney disease, who is still smoking cigarettes but is motivated to quit. There is a litany of advice I might give this person – quit smoking, take a statin, check your blood glucose, go see a foot doctor, an eye doctor, the list goes on. Without a prioritization scheme to help weigh the costs and benefits of each action against each other, this advice quickly becomes overwhelming and runs the risk of being ignored entirely. This is the power of the "next best action" approach in machine learning: tee up the most impactful and relevant care suggestions one at a time, or in related groupings.

Offering a plan that is personalized, value-based, and scalable to all members may sound daunting, but it can be done. The key is intelligent recommendations based on focused data inputs across an employee population and using it to deliver better health outcomes. If done well, scaling personalization can also lead to better health equity, an area employers are increasingly called upon to help address, and a key tenet of diversity, equity, and inclusion efforts.

Proven

So much of health care and benefits boils down to behavior change. We're asking people to do something different, whether it's see a different provider, get a lab test, or change a habit, all in service of better health at lower costs. And let's face it, behavior change is hard. To be effective, employers need to give the right information to the right person, through the right communication channel, at the right time. Not only do we need to build plans and present care suggestions that are grounded in evidence, but we need to use evidence to drive how we approach cost-effective navigation of health benefits.  

In many instances, a lack of information delivered properly will lead to care avoidance altogether. A 2020 Policygenius survey on health insurance literacy taken during the COVID-19 pandemic found that around 1 in 4 Americans avoided care because they didn't know what their health plan covered. Employees on high deductible plans are more likely to avoid care, including necessary care that should be free under the plan, like immunizations. This is evidence that the status quo is not working and we should all challenge it, asking the hard questions and measuring the important outcomes to give ourselves a clear-eyed view of whether the structure we're providing is giving us the desired impact. 

We can study the impact of the words we use, to learn what member-facing content has the greatest impact. We can even build products that apply principles of behavioral economics, relying on proven concepts like the power of default settings in the provider search, to help guide people to the highest value within their plan offering. While there is still much to learn,‌ by measuring what matters and holding ourselves to a standard of evidence-based care navigation, we have the chance to do more of what works and less of what doesn't in a cycle of continuous improvement. 

Pleasing

Finally, care needs to be easy to navigate for the individual end user. The process of personalizing care will not work if it adds to the complexity, fragmentation, and inefficiency of the system. If the process of selecting care proves strenuous on the individual, they won't be motivated to take the steps they need to get care.

To curate a seamless user experience, care connectivity and interoperability needs to be handled on the back end. An effective member experience strategy combines digital innovation, grounded in easily navigated interfaces, with human support to help employees better understand and manage their care. Our health care system should be as easy to use as the internet, and that high-touch care management starts with a simple question around delighting members with an intuitive experience. 

It may still be early innings on some of this, but the flywheel is turning on the use of health care data to drive better value in the employer benefit space. When considering benefits design through the lens of the "Four P's," advisors, HR professionals and otherbenefits leaders can build a plan that puts value at the forefront, addressing the unique and varied needs of all their people in service of better health at lower costs. 

Dr. Ari Hoffman is the chief clinical officer and VP of population health at Collective Health.