Jackknife Instrumental Variable Estimation of the Effects of Gamma Knife Radiosurgery in the Treatment of Metastatic Brain Tumors

Abstract

I investigate the effect of Gamma Knife radiosurgery on health and cost outcomes in patients with metastatic brain tumors using a jackknife instrumental variable that measures physician propensity to use Gamma Knife. The instrumental variable regression is designed to attempt to control for unobserved endogeneity in the decision to treat using Gamma Knife and reflects the fact that physicians may observe patient characteristics not recorded in datasets. An understanding of the effects of Gamma Knife can help inform physicians and patients regarding optimal treatment plans and can improve efficiency in the healthcare system. Using the jackknife instrumental variable, I find evidence that Gamma Knife is associated with improved health outcomes and also increases costs. Specifically, treatment using Gamma Knife reduces the severity of a patient’s Risk of Mortality ranking by 42.0%, the Severity of Illness ranking by 48.6%, the length of hospitalization by 117%, and the probability of complications by 66.7%, while increasing costs by 66.8%. These results suggest that although Gamma Knife is not frequently used for treatment of brain tumors, it may constitute an effective treatment for certain patients.