Area-level variations in cancer care and outcomes

Publication Name: 
Medical Care
Publication Authors: 
Keating NL, Landrum MB, Lamont EB, et al.
HCP Authors:Nancy Keating MD, MPH, Mary Beth Landrum PhD, Elizabeth Lamont MD, MS, Barbara J. McNeil MD, PhD
Date of Publication: 
May 2012


BACKGROUND: Substantial regional variations in health-care spending exist across the United States; yet, care and outcomes are not better in higher-spending areas. Most studies have focused on care in fee-for-service Medicare; whether spillover effects exist in settings without financial incentives for more care is unknown. 

OBJECTIVE: We studied care for cancer patients in fee-for-service Medicare and the Veterans Health Administration (VA) to understand whether processes and outcomes of care vary with area-level Medicare spending. 

DESIGN: An observational study using logistic regression to assess care by area-level measures of Medicare spending. 

SUBJECTS: Patients with lung, colorectal, or prostate cancers diagnosed during 2001-2004 in Surveillance, Epidemiology, and End Results (SEER) areas or the VA. The SEER cohort included fee-for-service Medicare patients aged older than 65 years.  MEASURES: Recommended and preference-sensitive cancer care and mortality. 

RESULTS: In fee-for-service Medicare, higher-spending areas had higher rates of recommended care (curative surgery and adjuvant chemotherapy for early-stage non-small-cell lung cancer and chemotherapy for stage III colon cancer) and preference-sensitive care (chemotherapy for stage IV lung and colon cancer and primary treatment of local/regional prostate cancer) and had lower lung cancer mortality. In the VA, we observed minimal variation in care by area-level Medicare spending. 

DISCUSSION: Our findings suggest that intensity of care for Medicare beneficiaries is not driving variations in VA care, despite some overlap in physician networks. Although the Dartmouth Atlas work has been of unprecedented importance in demonstrating variations in Medicare spending, new measures may be needed to better understand variations in other populations.

(May 2012)

View the Article