Healthcare Analytics, Population Health Management, Healthcare Big Data

Population Health News

Leveraging Risk Stratification for Population Health Management

What do providers need to know about developing risk stratification competencies to bolster their population health management activities?

By Jennifer Bresnick

- Understanding a patient’s risk for developing new conditions – and a provider’s financial risk for treating those conditions – is the foundation for meaningful population health management, according to a new report from Chillmark Research.

risk stratification for population health management improvements

However, many providers do not have the data or the techniques to create a comprehensive portrait of total risk, which should include the social and behavioral determinants of health along with traditional clinical and claims data.

"Current risk models were largely developed for actuarial purposes,” explains lead author Jody Ranck.

“We now know claims and clinical datasets are not enough to effectively predict true risk for all high utilizers, and it is essential for future business success to incorporate additional measures to accurately manage the need for current and future services.”

Risk profiles that rely solely on data from electronic health records and claims may only account for ten percent of a patient’s actual probability of experiencing an expensive and serious outcome. 

READ MORE: How to Use Big Data for Tailored Population Health Management

Integrating additional data about socioeconomic status, pharmacy use, available community resources, educational status and capacity, and location-based factors such as transportation availability or access to healthy food choices, may account for another 70 percent of data that contributes to better health.

“Currently, there are only a handful of healthcare organizations and vendors tackling this problem, and those that do it well will have a significant competitive advantage in the coming years,” said Ranck.

The report stresses the importance of utilizing integrated, sophisticated, and holistic big data analytics to capture more than just a snapshot of risk at a single point in time. 

Flagging one emergency department visit for asthma within the last sixty days may heighten risk of a hospitalization by some degree, but three or more ED visits in the past six months coupled with several primary care visits and an air pollution warning in the area is an indication that the patient may have a worsening condition that isn’t being adequately addressed.

In order to create a total active risk profile, the report suggests that healthcare providers work to personalize interventions for patients by combining improved analytics with tailored population health management strategies.

READ MORE: Value-Based Care Requires Good Big Data, Better Communication

Identify factors that increase risk

The industry’s growing population health management experience has helped to flag certain patient characteristics that are widely accepted to be instrumental in predicting risk.  These may include medication adherence history, the number of medications prescribed to a patient, emergency department utilization, high utilization rates for other services, inadequate monitoring and follow-up participation, previous non-compliance with treatment regimens, lower levels of patient engagement and education, and a lack of community infrastructure or caregiver involvement.

They may also include basic socioeconomic circumstances, such as race, income level, and ethnicity, that hinder access to care or produce deeply entrenched cultural responses to certain lifestyle choices.

When developing population health management interventions for particular disease states, providers should analyze their own potential patient cohorts to discover which of these factors may have the most impact on the desired outcome.

Choose the right patients for participation

READ MORE: 5 Test Cases to Prove the Value of Population Health Management

Not all patients will respond to a single population health management activity in the same way.  Extremely complex patients, for example, may not benefit much from a basic educational program about choosing the right foods for managing kidney disease.

“Risk stratification will need to provide insights into which patients will respond to which interventions at the right time in order to provide value to healthcare organizations,” the report says.  “Risk stratification efforts must be coupled to both the downstream outcomes desired by the organization and the actual interventions the organization offers.”

Providers must start with a solid understanding of the value and scope of the programs they can offer.  If the organization does not have the ability to deliver mental healthcare services in-house, they may need to partner with another provider to bring management programs to patients with significant behavioral health problems.

Organizations must also be able to separate “management” from “prevention.”  A pre-diabetic patient may be a good match for a weight loss program that would help them forestall the development of full-blown diabetes, but would not receive much value from an educational session focusing on what to do after a diabetes diagnosis.

Assess gaps in the care continuum

Poor care coordination, paired with variable patient behaviors, may lead to gaps in the care continuum that have serious consequences for managing a chronic disease or reducing utilization. 

Healthcare organizations must take a dual approach to this problem.  Not only do they have to assess where they are falling short in accessibility or coordination, but they must be able to flag patients with utilization rates above the target threshold and identify what specific actions could help narrow the divide between the care they are receiving and the care they need.

While the top ten percent of complex patients are responsible for the majority of healthcare spending, and it is important to address those situations, there may be additional opportunities to create impacts on outcomes and costs among the middle tier of utilizers: those who use the emergency room instead of an urgent care center, for example, or patients who do not regularly refill their medications and so end up in a crisis state.

These problems may be ameliorated by educating patients about the availability of after-hours care centers, helping them switch to a mail-order pharmacy service that allows them to easily access their prescriptions, or adding routine check-ups to catch developing symptoms before they worsen into an emergency.

Plugging these gaps in the care continuum requires a thorough understanding of what produces risk for a particular patient and how receptive that individual is to interventions.  It will also require providers to access alerts and data from their business partners, such as admission, discharge, and transfer (ADT) notifications from a local hospital or care summaries from specialists.

Develop patient engagement and competency to participate

No population health management intervention can be successful unless the patient wants it to be.  Unfortunately, not all patients – and not all providers – are able to work together to change unhealthy behaviors, connect with appropriate resources, commit to attending appointments and sessions, or raise health literacy levels.

Providers must remember that patient engagement “requires meeting patients where they are, rather than providing cookie-cutter solutions that patients must conform to for treatment,” the report states. “Healthcare stands much to learn from other fields in how to communicate interventions based on education, social context, language, gender, and other relevant data.”

While some patients may be overjoyed to learn they can access their electronic health record data from a patient portal, others might balk at the idea of having to reveal that they have a very low level of computer literacy.

The growing popularity of wearables and Internet of Things devices might make it easier for some patients to track their diet and exercise, but may not produce much clinical benefit for a patient with four chronic diseases who is also struggling to find transportation to their appointments each week.

A comprehensive risk stratification model will take into account a patient’s engagement profile and willingness to participate in healthy activities as well as their clinical state.

Commit to continuous improvement with the use of big data

“Risk stratification is understood as a learning tool that enables providers to generate insights that are continually fed back into the cycle to modify approaches based on what the data demonstrates is working,” says the report.

While all healthcare organizations can benefit from adjusting their activities based on lessons learned from past experiences, it may be even more important for providers participating in value-based care arrangements.

Cutting costs while boosting quality and patient outcomes is vital for success under accountable care organization contracts or other risk-based reimbursement methods, which means providers must be able to use their big data to assess their own performance.

“The cultural transition by clinicians to a 1`more risk-based form of contracting will require more attention and training to optimize use of analytics for care management and risk stratification, which will need to be part of an overall data governance strategy,” Chillmark states.

“In the short term, we expect the focus to be on low-hanging fruit for value-based care contracts as they now stand, which translates into a greater focus on tools that offer insights on trending vitals. The next level would be to integrate behavioral health data from EHRs.”

Socioeconomic data and community information should be on the agenda after that, but the report notes that “social data is likely to take a back seat to integration of clinical and behavioral data followed by more extensive use of patient-generated data in the next two to three years.”

“Depending on the success of the leading vendors that are providing social data analysis…and corresponding ability to demonstrate an ROI from social data, we expect more interest in this area in five years or more.”


Join 25,000 of your peers

Register for free to get access to all our articles, webcasts, white papers and exclusive interviews.

Our privacy policy

no, thanks