Overdiagnosis and the Will Rogers Effect: Bias in Risk Adjustment

• Activity Number: 350
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Daniel Gottlieb*+ and Jason Sutherland
• Companies: The Dartmouth Institute for Health Policy and Clinical Practice and Centre for Health Services and Policy Research
• Keywords: risk adjustment ; bias ; health policy

Abstract

To compare regions health systems' outcomes, risk adjustment for chronic disease is an accepted methodology. This approach assumes consistent coding. We test whether differential coding is occurring and whether it is associated with an area's average health care spending. Meth: We divide the nation into 306 Regions and assign each to a quintile of spending intensity (Q1:lowest). We then determine mortality and disease prevalence and show their relationship with intensity. We then run models of people coded with and without disease. Overdiagnosis will result in both the sick and healthy groups having lower mortality rates. Res: Chronic disease prevalence increases with area level spending (PVD Q1:6.4% Q5:11.1%), but is unassociated with crude or adjusted mortality (Q1:5.0 Q5:5.0%). When stratifying by presence of disease, higher cost areas have ~5% lower mortality among both groups.