ABSTRACT
Human immunodeficiency virus (HIV) studies frequently emphasize the quantification of HIV ribonucleic acid (HIV RNA) copy numbers per unit volume of biospecimens. Viral load assay sensitivity has improved substantially with time, and reaching a nondetectable level is often used as a surrogate endpoint in clinical studies of HIV infection. However, the issue of left censoring due to values below detection limits continues to pose challenges to the analysis of continuous HIV RNA data. Laboratory assay nondetectables are pervasive enough in multiple areas of science to have spawned statistical research, dissertations in biostatistics or statistics, and a myriad of journal articles. To deal with bivariate problems such as those arising in HIV or environmental epidemiology, considerable subsequent attention was paid to parametric approaches for estimating crude correlation coefficients in addition to means and variances when one or both variables are subject to nondetectables. Laboratory nondetectables are a common occurrence in epidemiologic and clinical health-related research.
