New Population Health Decision System For ACOs Uncovers The Hidden Risks

By Saurabh Swarup, Global Solution Leader for Provider Analytics, Dell Healthcare Services

One of the biggest challenges for Accountable Care Organizations is effective deployment of patient care resources. To succeed requires an in-depth understanding of patient risk, which is often difficult to assess. The reason for this is that most risk analysis systems rely on claims data, which can only tell who, in the past, has been a high-use patient. What that data can’t do is tell you who is on the verge of developing a serious condition.

But that gap is closing. Baystate Health, a large community-based integrated health system in Western Massachusetts, is developing a new Population Health Risk Analysis system that uses clinical and socioeconomic data to give physicians a much better look at where resources are needed most. It is being developed as an innovation project through TechSpring, the Baystate Health Technology Innovation Center, led by CIO Joel Vengco.

A Next Generation ACO

Baystate is a partner member of the Pioneer Valley ACO, a regional entity which brings groups of doctors, hospitals, and other partners together to provide coordinated, high quality, cost-effective care to patients in the community. Pioneer Valley is one of the 21 ACOs that the Center for Medicare & Medicaid Services Innovation has designated as a participant in the Next Generation ACO program. These are ACOs that have significant experience coordinating care for populations of patients through initiatives like the Medicare Shared Savings Program and the Pioneer ACO Model. Building on experience from these previous initiatives, CMS is partnering with experienced ACOs whose provider groups are ready to assume higher levels of financial risk and reward.

That means that Baystate and the other providers in the Pioneer Valley ACO have a big stake in knowing who is at risk and deploying resources where they can be most effective.

The community the ACO serves is made up of people of all ages and many ethnicities, with widely varying health needs, from chronic disease management to acute care and preventive and well care. The community also has a large population of people with socioeconomic risk factors, such as low income, a home location that is far from a primary care facility, lack of family support and communication issues due to language. Often these patients lack money for prescriptions and co-pays, don’t have a medical home and may go to the closest ER to get sporadic care.

With a diverse population and a wide variety of risk factors, accurate identification and stratification are important, along with identifying gaps in care.

Predicting Future Risks, Not Just Reporting Past Claims

A major difference between most risk prediction systems and Baystate’s new system is the source of the data. Other systems use claims history to identify patients who need extra attention, which is helpful for many patients but misses other patients who could benefit from preventive services.

Baystate’s tool incorporates clinical data from physician and hospital EHRs (such as vital signs, labs, medications, procedures) as well as socio-economic data to identify patients with rising risk. These are the patients with clinical markers and socioeconomic risk factors that indicate a high probability of developing a chronic disease and who need medical intervention. Because these patients have not yet needed frequent or expensive services, claims data alone won’t identify them as having high risk and they might not receive the preventive services they need.

For example, let’s say you have a patient who is overweight, has high cholesterol and glucose levels that are on the high end of normal. The patient has not been a frequent user of healthcare services, so a claims-based risk assessment would not identify the person as high risk. This patient, and others with similar issues, would be missed by analytics based primarily on claims data. They are a hidden cohort of patients who harbor significant risk of future problems and who could benefit from preventive services.

The new system considers clinical and socio-economic data in addition to claims data. The clinical information (such as weight, lab values and family history of chronic disease) and socio-economic factors might flag many of these patients as having rising risk. This gives the caregiver an opportunity to intervene before the kind of health deterioration that results in high medical costs and increased suffering.

I’ve talked with many physicians who would welcome the opportunity this risk analysis system offers to make a real difference in a patient’s health and, perhaps prevent a chronic condition. For decades, family physicians have labored to hold back the tsunami of chronic disease, with little actual progress. But with remote monitoring and health coaching, physicians can give these patients a deeper understanding of their disease and risks and help them change their daily lives in ways that profoundly improve their health. There is actual clinical evidence that this combination of coaching and monitoring can change the course of chronic disease. And the new tool is critical to identifying those patients at a time that offers the greatest opportunity for Baystate physicians to help them live a healthier life as well as to lower the ACO’s financial risk.

As with any project that Baystate undertakes that involves patient data, attention to patient privacy and compliance with HIPAA protections is always of critical importance.

Developed By Physicians, For Physicians

It is also important to note that the new risk analysis tool is being developed by physicians for physicians, with technical assistance from analytics and computer experts. Dr. Neil Kudler and Dr. Nick Kashey have worked with Dell in this pilot to pull key clinical data from the EMR and billing systems and to create a user interface that fits their needs. This physician leadership has been critical to ensuring that the tool is easy to use and fits within the physician’s established work flow. That’s an important factor, because physicians’ time is already so burdened that if the tool requires them to go outside of their established routines, it is less likely to be used.

The project is now in the final stage of TechSpring’s innovation process focused on quantitative assessment of how much money can be saved — and how much suffering avoided. So far, test runs of the tool have identified numerous of these rising risk patients and provided actionable data that can be integrated with the care management process. Overall, it is clear that there is significant value in such rising risk analytics that use clinical and socio-economic data.