Is Patient Activation the Answer? Engaged Patients Could Yield Lower Costs for Hospitals

Accountable Care Organizations (ACOs) are growing rapidly in the United States because of new pay-for-performance incentives under the Affordable Care Act. These provider networks currently cover more than 28 million patients across the country, whereby they agree to cover a set number of patients for a fixed cost per year. One of the central tenets of the ACO model involves providers agreeing to take on some level of financial risk for the patients in the population they choose to manage (e.g., patients within certain geographic areas, patients with certain conditions), because the health systems are no longer reimbursed for each individual service that they provide. Thus, it has become critical for providers to have a better understanding of which patients are at high risk for excessive utilization, since those patients could have a large impact on a health system’s bottom line.

Historically, health systems have tried to identify high-risk patients using data on past utilization and what current health conditions a given patient might have (e.g., diabetes, hypertension). A variety of algorithms have been used to combine this data into scores or rankings that hospitals can use to identify patients. However, on average, many of these algorithms yield mixed results, perhaps because trying to use past utilization to predict future utilization is not always effective.

There has been increased focus in recent years on trying to better understand the extent to which patients take charge in managing their own conditions. This concept, known as patient activation, suggests that risk prediction models will be incomplete if they do not also take into account whether a given patient is able and willing to take steps going forward toward better self-management. Even if a hospital can identify potentially risky patients with utilization data alone, the health system will be less effective at managing those patients without understanding patient activation levels.

In a recent study in Health Affairs, Judith Hibbard et al. examine whether including measures of patient activation improves the ability of existing risk prediction models to identify high-risk patients. The authors use a survey-based scale called the Patient Activation Measure (PAM) that groups patients along a four-point leveling scale based on how activated patients are. Level 1 represents patients who have the least understanding of the role they play in determining their health, and Level 4 represents patients who are proactive in managing their own conditions. The authors studied patients within Fairview Health Services, a Pioneer ACO in Minnesota, which collects PAM scores for its patients within its electronic medical record system.

The authors first used a standard risk-prediction model to isolate the top 15 percent of patients within the Fairview system who had the highest risk for future costs. The team then developed a new model that would integrate this risk score, demographic characteristics, and the PAM measures to predict whether or not these high-risk patients had a costly hospital visit in a given year (2012, 2013, or 2014).

Hibbard et al. found that patients with lower levels of activation were significantly more likely to have Emergency Department (ED) visits and hospitalizations than those with higher levels of activation. Level 1 patients had over 25 percent higher odds of ED visits and hospitalizations than did Level 4 patients. Level 1 patients also had hospitalization costs that were 38 percent higher in 2012 and 29 percent higher in 2014 than did patients in Level 4, and patterns for ED visits were nearly identical. Furthermore, for nearly all three years of the study, ED and hospital costs decreased as the level of activation increased, and those at Level 1 generally had costs and hospitalization rates that were significantly higher than patients at any other level of activation. Overall, the study finds that patient activation is a significant factor underlying utilization, resulting in an average cost differential between Level 1 and Level 4 patients of more than $2,000 per patient.

This study provides evidence that patient activation data can help create a fuller picture of what makes a particular patient “high-risk.” As ACO models become more prevalent and aim to take on larger patient populations, accurately predicting patient risk will become vital to the financial success of health systems. While patient activation is still one component of understanding risk, this study provides evidence that in order to survive in the ACO universe, health systems will need to continue to build capacity to measure and improve patient activation.

Article Source: Hibbard, Judith, Jessica Greene, Rebecca Sacks, Valerie Overton, and Carmen Parrotta. “Adding A Measure Of Patient Self-Management Capability To Risk Assessment Can Improve Prediction Of High Costs.” Health Affairs, 2016.

Featured photo: cc/(MJFelt, photo ID: 58321222, from iStock by Getty Images)

Matthew Green
Matthew Green is a staff writer for the Chicago Policy Review. He is interested in the private health insurance market and healthcare payment model reform.

Comments are closed.