This Article Full Text (PDF) References Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to Saved Citations Download to citation manager Request Permissions Request Reprints Add to My Marked Citations Citing Articles Citing Articles via Scopus Google Scholar Articles by Cole, S. R Articles by Lei Nie PubMed PubMed Citation Articles by Cole, S. R Articles by Lei Nie, Social Bookmarking                         What's this?

Nonparametric estimator of relative time with application to the Acyclovir Prevention Trial

Department of Epidemiology, UNC at Chapel Hill, Chapel Hill, NC 27599-7435, USA, cole{at}unc.edu

Haitao Chu

Department of Biostatistics, UNC at Chapel Hill, Chapel Hill, NC 27599, USA

Lei Nie

Office of Biometrics, CDER/OTS/FDA, Silver Spring, MD 20993, USA

Background Relative hazard is a central measure of association in randomized clinical trials. Relative time (RT) is a competing measure that is rarely used.

PurposeWe describe a simple area-based nonparametric estimator of RT and illustrate its use in the Acyclovir Prevention Trial.

Methods Let Q _x(p) be the quantile function for the xth treatment group, defined as the time by which p% of the treatment group experience the event, and p _x be the maximum event proportion observed. Our consistent estimator is defined as the ratio of the integrals of Q₁(p) and Q₀ (p) with integration over 0 to p, where p =min(p₁, p ₀). Confidence limits (CL) are provided by bootstrap.

Results A total of 703 immunocompetent adult men and women (54% male, 79% Caucasian, median age 49 years) with a history of ocular herpes simplex virus (HSV) were enrolled in 1992—1996, randomized to acyclovir or placebo, followed for up to 1 year for the 1st episode of ocular HSV, and 170 events were confirmed by a study-certified ophthalmologist using slit-lamp biomicroscopy. The nonparametric RT comparing acyclovir use with nonuse was 2.6 (bootstrap 95% CL: 1.6, 4.2). For comparison, the best-fitting parametric model was the lognormal (RT = 2.5; 95% CL: 1.5, 3.9). In limited simulations, the average proposed estimate of RT was similar to the true RT with a relative root mean squared error of 1.13 compared to a correctly specified parametric (lognormal) model.

Limitations An analytical variance estimator for the proposed RT is lacking. Also, more examples and more extensive simulations are warranted.

Conclusions Similar to Cox’s relative hazard estimator, the proposed RT does not assume the data are generated from a particular distribution. RTs should be more widely used as a measure of association in clinical trials. Clinical Trials 2009; 6: 320—328. http://ctj.sagepub.com

Clinical Trials, Vol. 6, No. 4, 320-328 (2009)
DOI: 10.1177/1740774509338231


CiteULike    Complore    Connotea    Del.icio.us    Digg    Reddit    Technorati    Twitter    What's this?