Abstract:
|
The incremental cost-effectiveness ratio (ICER) is used to compare competing interventions for the same goal. It is a function of the expected life span (ELS) and the unit cost of treatment over the life span for each intervention compared. We develop an estimator of ICER for use in both observational cohort studies and randomized clinical trials (RCTs), that allows adjusting for any number of covariates which are determinants of the outcome of interest. These covariates are thus modifiers of the ICER even when they are not confounders of the estimated intervention effect. We also develop an estimator of the restricted ELS, that is, the ELS between two time landmarks, using a semi-parametric Cox regression model. In addition to its utility in allowing for effect modification by outcome risk factors, the proposed approach is applicable under left truncation and when the proportional hazards assumption does not apply. The method is illustrated through an analysis of the cost-effectiveness of switching from 1st to 2nd line antiretroviral drugs among 2nd line eligible patients in Dar es Salaam, Tanzania. A user-friendly SAS macro is available for implementing the method.
|