Over 500 Total Lots Up For Auction at Six Locations - Jutland 09/12, CA 09/19, TX 09/20, CA 09/23, NY 09/24, MI 10/03

O modelo do computador prediz pacientes “maus”

por Brendon Nafziger, DOTmed News Associate Editor | October 11, 2010
A drug benefits management company says its computer models can predict up to a year in advance whether a patient will stop taking his meds, thereby helping save the health system non-adherence-related costs.

Express Scripts said Monday it had finished testing the models that predict whether patients will follow doctors' orders for diabetes, high blood pressure and high cholesterol medications.

The aim is to intervene by contacting the patients before their non-compliance threatens their health and therefore raises health care costs. An April report by Express Scripts said $106 billion was wasted every year from non-adherence-related medical costs.

"The problem of non-adherence isn't new - it's easy to walk through a hospital and identify people who would not be there if they had simply taken their medications," Express Scripts' chief medical officer Dr. Steven Miller said in prepared remarks. "Our new predictive models allow us to do something that wasn't possible before: better identify those patients before they run into trouble, and tailor practical, patient-centric solutions that target the specific factors that put them at-risk for non-adherence."

Factors that raise the risk of skipping meds include having minor children at home and co-payments greater than 50 cents per daily dose, according to the company. Age is also a factor.

"Older individuals are more likely to stay on their medications than younger folks," David Whitrap, a spokesman for the St. Louis-based company, told DOTmed News.

Whitrap said the computer model is around 80 percent accurate in predicting the top 10 percent most likely to adhere, and the bottom 10 percent least likely to adhere to medication schedules. He said previous models in the industry and academia were in the 60 percent range.

For diabetes patients, the model was most accurate: achieving 98 percent prediction success for the least adherent group, Whitrap said.

"With that degree of confidence, we were able to intervene with the folks who need it most and concentrate it in that direction, and not pester folks likely to stay on medication already," he said.

As for how successful the interventions were in increasing compliance, the company said it's too early to give data.

"We know they are impactful, but we don't have numbers yet to report," Whitrap said. "But we expect to be able to release those in the next few months."

The program is running now on a pilot basis, but Whitrap expects to roll it out across the company's entire client base over the next year.

Express Scripts isn't alone in this field. CVS Caremark, one of the nation's biggest pharmacy chains, is experimenting with its own non-compliance prediction system that it also intends to roll out soon.