Modern Healthcare: Are predictive analytics essential for 'population health 2.0'?

Experts think so

Hoping to improve patient outcomes and lower costs, more health systems are investing in predictive analytics. Can the technology deliver on its promise? Modern Healthcare's Melanie Evans examines the trend.

Advocate Health Care: A model to follow?

Evans notes that Medicare's 2012 decision to begin penalizing hospitals for readmissions set providers scrambling to "identify and head off potential repeat visitors." (Those penalties have since proved costly; Medicare has collected more than $500 million from hospitals to date.)

Seeking to simultaneously avoid penalties and raise the quality of care, many organizations elected to invest in analytics, either by partnering with external firms or growing a software solution themselves.

At Illinois-based Advocate Health Care, eight of the system's 11 hospitals now rely on a "big-data" algorithm that is 20% more accurate than other predictive algorithms on the market, Advocate officials say. The algorithm combines data from patients' medical records, claims, demographic, laboratory results, pharmacy use, and self-description of their health status.

Buoyed by its initial success, Advocate plans to launch a second predictive-analytics initiative this year that evaluates all patients receiving care from in-network physicians. The tool will sort patients based on the complexity of their conditions and aim to identify patients that would make the best candidates for interventions to prevent disease, allow patients to more feasibly manage health conditions outside a hospital setting, and prevent readmissions.

Analytics offer potential, but many challenges in making it real

A growing number of organizations—from insurers like WellPoint to technology firms—are developing systems to help integrate hospitals' clinical information into predictive models, Evans writes.

The seven factors that could predict readmissions

However, Modern Healthcare notes that hospitals have faced technological hurdles when trying to implement predictive analytics. Some systems have fallen short because they solely rely on claims data, which can include redundant information and be time-intensive to navigate. Other efforts have been foiled by incomplete information in patients' EHRs or data residing in multiple records.

Meanwhile, many predictive analytic systems on the market have focused on targeting patients with the potential to be most costly to the hospital, or the 5% that account for 25% of spending. According to Rishi Sikka, Advocate's senior VP of clinical transformation, targeting these expensive patients is the "first generation of population health," but using more sophisticated algorithms to determine the medical needs of an array of patients is "population health 2.0." And that sets up a tall order for new analytic systems, given the difficulty in unlocking this information.

According to David Nash, dean of the Thomas Jefferson University School of Population Health, "Understanding as much as you can about all aspects of your patient, not just their disease, but their social setting, their history of utilization, their risk for hospitalization, that's big data in health care," adding, "The more you understand, the more efficiently you can deploy resources" (Evans, Modern Healthcare, 7/12 [subscription required]).

Improving sepsis outcomes with Crimson

We mined the Crimson cohort for best practices and clinical trends and identified nine areas where hospitals can change care processes to make the most impact on outcomes. Our infographic outlines process improvement steps and corresponding data points for performance monitoring—and how Crimson can help.


Next in the Daily Briefing

The latest industry transitions

Read now