Impactability is the next step for predictive analytics

As health organizations shift from reactive to preventative care, many decide to focus on analyzing patients’ risk of succumbing to certain conditions. Organizations typically assign a “risk score” for each patient to allow clinicians to easily assess which patients are at a higher risk of an event that may require intervention. While such initiatives typically yield positive outcomes, it’s important to keep in mind that these are only the first steps in what is still the early days of predictive healthcare.

While risk scores are effective at assessing what patients are most likely to have an event or fall ill to a condition, they fail to take into account the many factors that can contribute to a patient’s likelihood of being affected by care. For example, while two patients may be equally likely to visit the emergency room within two weeks, throwing an equal amount of resources and treatment at the patients almost never equates to an equally effective outcome. Factors such as age, personal health history, lifestyle, and level of compliance all influence the effectiveness of the treatments on those patients.

However, building atop the solid foundations of assigning risk scores, organizations can begin to shift focus away from measuring risk and toward determining impactability. Simply put, impactability is the extent to which an intervention can affect the health and costs of a patient.

Thanks to ongoing technological developments, patient data is more thorough and readily available. For example, sophisticated modeling software now allows healthcare organizations to predict patient behaviors and conditions, identifying if a patient uses certain healthcare services too much or too little and applying that knowledge to determine how effective treating that patient is. By focusing on how effective treatment can be on a given patient, clinicians can better determine the proper amount of resources to devote to a patient, which can cut costs and increase ROI.

Determining Impactability

A study by The Milbank Quarterly highlighted three methods of determining impactability that organizations should strive to identify.

The first method is identifying patients with diseases that are particularly amenable to preventative care, which requires excluding those patients with the very highest risk and giving priority to ambulatory care-sensitive conditions, as well as giving priority to patients based on the gaps in their care—defined as “an observed difference between the optimal care and the care received.”

The second method is to exclude patients who are least likely to respond to preventative care, such as those who have demonstrated previous noncompliance with treatment or have characteristics that may make them less compliant, patients who are very similar to previous patients who did not successfully manage a preventative program, and patients at high risk of disenrolling in a disease management program.

The third method is to tailor the form of preventative care to each patient‘s characteristics, such as determining the best medium with which to contact the patient, determining the type of preventative care best suited to each patient, and utilizing incentives to to encourage patients to engage with their care plan.

Assigning an Impactability Score

For impactability to be properly understood, however, it should ideally be put into a score similar to a risk score; assigning each patient a number that portrays at a glance how much change can be expected as a result of intervention. One organization that has found success with this method is Community Care of North Carolina (CCNC), who developed an impactability score for prioritizing patients for complex care management. Here, CCNC found that “by examining an individual’s pattern of preventable spend relative to clinically similar patients, [it was] able to identify pockets of opportunity that would have been missed with traditional approaches,” and developed a score that relates directly to potential cost savings after intervention.

Through its initiative, CCNC found only a 53% overlap between the top 5,000 highest impactable patients and the top 5,000 patients with highest risk of inpatient admission. This translated to 2,327 highly-impactable patients that would have been missed with traditional risk-based methods and nearly $7,500,000 estimated savings over six months for the missed population.

Given the successful projections of this initiative, leaders at CCNC have applied a similar score to improve the organization’s low birth weight rate among Medicaid patients. Though this initiative is ongoing, the organization has already estimated that CCNC could prevent as many as 700 low-weight births during the course of the year. While the long-term effects of this are unknown, it’s expected that it will save the state millions of dollars, as it prevents not only the initial NICU stay but also gets children off to a healthier start, as their brains and organs will be more developed and as such the children will require fewer early intervention services during their developing years.

As preventative healthcare continues to gain momentum, it is important for healthcare organizations to stay on top of the latest trends and not simply become content with what has worked in the past. Though the concept of assigning risk scores may seem like an endpoint, it is only the beginning in the long journey through the continually shifting field of predictive analytics. While the next step beyond impactability may not yet be known, it is nonetheless an important step for the healthcare industry to take in order to be prepared for what comes next.