Many health care systems, including that of the Department of Veterans
Affairs (VA), are facing dramatic changes as they adapt to state-leve
l reform legislation and move into managed care environments. As new c
osting systems are designed and as researchers seek to conduct cost-ef
fectiveness studies to guide health policy, it is critical to examine
the assumed validity of the measurement procedures on which these cost
ing systems are based. The foundation for VA health care costing is th
e cost distribution report, which is created by combining data on expe
nditures and workload with local service chiefs' estimates of program
staffing and resource use. To evaluate the accuracy of the cost distri
bution report, the authors compared data from the report to cost data
obtained in three multisite studies. Substantial differences were foun
d between the cost distribution report and these independent sources,
suggesting that researchers should not place uncritical reliance on th
e cost distribution report. The accuracy of costing data in informatio
n systems being implemented by VA will be limited to the extent they r
ely on the existing VA cost distribution system, Several strategies ca
n be used in studies with a cost component to compensate for imperfect
cost distribution systems.