摘要：This paper investigates the relative efficiency of cancer treatment in Nova Scotia using data envelopment analysis (DEA) and patients as the base decision-making unit. Using a unique data set, DEA treatment efficiency scores are included as a variable in a second stage (ordered) probit model measuring the likelihood of patient survival. The estimated efficiency scores generally have positive coefficients in the econometric models, suggesting that certain treatments increase the probability of survival. But more importantly, we find that inter-hospital or inter-regional differences in patient population (e.g. severity of illness, other health complications and living conditions) are also helpful in determining health care outcomes. We conclude that health care policy which ignores inherent differences in patient population is not acceptable as a basis for public decision-making.
关键词：Cancer treatment; efficiency; probit estimation; health care policy.
