On the use of survival analysis techniques to estimate medical care costs

Survival analysis techniques are inappropriate methods in estimating the cost of health care. The use of the technique can result in dependent censoring that will increase the cost estimates derived by the Kaplan-Meier method and prejudice the figures used in multivariate Cox regressions. Survival analysis, which tends to make Kaplan-Meier-derived costing hard to interpret when survival is over the maximal censoring time, could also violate basic assumptions in proportional hazards with the use of Cox regression models on medical cost data.