Huge health expenditures have caused an unprecedented financial burden on the National Health Insurance in Taiwan. Understanding the health care seeking behavior of the population is crucial for re-designing the payment system. We proposed to use a two-phase model to examine the behavior of the population. Phase I model estimated the logarithm of odds ratio for a person going out of town for common cold controlling for demographics. Regional variation was treated as random spatial effects. Phase II model explored the regional effects using spatial features. The models were applied to 2001 National Health Interview Survey in Taiwan. In phase I, we found younger age (< 45 yrs), lower education level were more likely to go out of town for medical care. In phase II, we found the number of medical institutes and density of paved road were negatively associated with going out of town for care.
