Geographic Variation within the Military Health System

Document Type


Publication Date



Research Objective: This study sought to quantify variation in utilization of six surgical procedures and in per capita cost using system-defined geographic regions within the Military Health System (MHS).

Study Design: A retrospective cohort approach was used to analyze rates of utilization (hip replacement, knee replacement, coronary artery bypass graft, prostatectomy, and C-section) and per capita cost using MHS-defined regions (catchment areas) of enrollee residence as the unit of analysis. Areas where fewer than five procedures were performed were excluded. Since age and gender have been found to account for very little of geographic variation, unadjusted utilization rates and per capita costs were used.

Population Studied: This analysis studied 3.6M adult Tricare Prime Enrollees (including Active Duty service members, retirees, and their dependents) for the period for Fiscal Years 2007 and 2010 living in the United States. Tricare Prime is a HMO-like option for beneficiaries of the MHS. Prime enrollees are assigned a primary care manager who directs patient care, meaning a greater level of control should be in place for these enrollees.

Principal Findings: Variation, as measured by the coefficient of variation (CoV), was generally high for both cost and utilization. Per capita cost CoV was .29 for 2010 and .28 for 2007. By comparison, the CoV for Medicare per capita cost for Health Referral Regions from the Dartmouth Atlas was 15% for 2010. Utilization CoVs for 2010 ranged from .26 for C-sections (as a percentage of live births) to .48 for prostatectomies, and utilization CoVs for 2007 ranged from .24 for C-sections to .54 for knee surgeries. Procedures with lower rates generally had greater variation. A low but inverse correlation (-0.28) was found between the aggregate amount of care received in military facilities (versus private sector care) and overall utilization for back surgeries for a given catchment area.

Conclusions: Organized health systems such as the MHS might be expected to exhibit less variation than that documented for either Medicare or commercial insurance beneficiaries. These findings contradict this hypothesis, suggesting that other factors may be affecting to what extent variation occurs. Improved communication among military treatment facilities and a recently implemented ‘enhanced multi-service market’ strategy may help reduce variation. Moreover, design of managed care contracts for private sector care may offer a venue to discourage unwarranted care system-wide. Future analysis should confirm the role of age, gender, and race in this variation.

Implications for Policy or Practice: The Military Health System appears to be subject to the same cost pressures as the greater U.S. health system. Quantifying variation is an essential first step in reducing unwarranted variation in the provision of health care. Understanding the underlying mechanisms contributing to the findings of this study would assist not only leaders of the MHS, but also policy-makers for the US health system to determine if and where unwarranted care is being provided within the system.


Academy Health Research Annual Conference (ARM)


Minneapolis, MN