The introduction of potent antiretroviral therapies for treatment of HIV infection typically results in a dramatic reduction in plasma HIV RNA concentration, often to levels undetectable by current measurement practices. However, although a high proportion of patients achieve 'undetectability', many then experience a return to a state of detectability at a later date. As evaluation of virologic response provides a useful measure of therapy efficacy, it is of interest to estimate the proportions of cases with undetectable viral load over time following commencement of treatment. These proportions depend on the rates of transition from detectability to undetectability and subsequent return to detectability, and may be related to covariates or risk factors, possibly differing in both transitions. We consider construction of detectability profiles as estimates of these proportions, based on parametric modelling of the component survival distributions. The method is applied to an examination of the effects of baseline CD4 T-cell lymphocyte counts on virologic response to therapy amongst patients of the Western Australian HIV Cohort Study.
Copyright 2003 John Wiley & Sons, Ltd.