In a working paper on modernizing healthcare, UnitedHealth Group’s Center for Health Reform and Modernization is proposing the use of predictive modeling software, particularly in Medicare and Medicaid programs, as a possible solution to both healthcare fraud and preventable hospitalizations.
The National Health Care Anti-Fraud Association conservatively estimates that about 3 percent of U.S. healthcare spending is lost to fraud or payment and billing errors. The association also estimates that about 70 percent of payers use some form of anti-fraud system, with many still using the “pay and chase” methodology.
[See also: Healthcare fraud recoveries set record in 2012.]
Minnetonka, Minnesota-based UnitedHealth, one of the largest managed care companies and the parent company of the healthcare technology firm Optum, said in its working paper that payers are starting to embrace predictive analytics for fraud analysis, with Medicare and Medicaid increasingly adopting the pre-claims adjudication process used by commercial insurers.
The Centers for Medicare & Medicaid Services has a new fraud tracking system which uses predictive modeling. The system is currently being integrated with CMS' payment-processing system. Predictive modeling systems can pre-score claims prior to payment and identify billing irregularities.
As part of UnitedHealth’s broader proposals for modernizing Medicare and Medicaid, the company’s think tank said that predictive modeling could also go a long way to helping reduce repeated hospitalizations and for improving continuity of care.
Already used in Medicaid and private insurance and also being used in Washington state’s managed fee-for-service dual eligible demonstration projects, predictive modeling can identify patients at risk for hospitalization and patients “who might benefit from higher levels of care coordination or other interventions,” UnitedHealth’s report said.
“Based on our experience operating Medicaid managed care plans, those programs are most effective when they deploy critical interventions such as comprehensive care plan development, ongoing care coordination, home visits, management of high-risk patients and case management of care transitions and discharges to prevent hospital readmissions.” Predictive analytics, the report continued, “enables those services to be targeted most effectively, and integrating proven delivery reform models such as patient-centered medical homes into plan services also improves care for enrollees.”
Predictive modelling fits in with the type of community-based care models being used in several state Medicare-Medicaid dual eligible demonstrations, and UnitedHealth said it’s particularly useful in helping people with chronic health conditions. Care managers focus on transitions of care – between hospitals, family caregivers, nursing facilities, primary care doctors and community services – and they can also engage patients and help them manage or improve their conditions.
According to UnitedHealth’s report, “Bringing this type of targeted intervention to the Medicare fee-for-service population would require establishing a mechanism to identify the individuals who might benefit most and then to engage them through outreach programs. Medicare could develop a reimbursement approach for the kinds of nurse manager, care coordinator, and community health worker services provided through the program — for example, by paying a service fee to entities that engage those health professionals.”