osDesign: AnRPackage for the Analysis, Evaluation, and Design of Two-Phase and Case-Control Studies

Abstract

The two-phase design has recently received attention in the statistical literature as an extension to the traditional case-control study for settings where a predictor of interest is rare or subject to missclassification. Despite a thorough methodological treatment and the potential for substantial efficiency gains, the two-phase design has not been widely adopted. This may be due, in part, to a lack of general-purpose, readily-available software. The osDesign package for R provides a suite of functions for analyzing data from a two-phase and/or case-control design, as well as evaluating operating characteristics, including bias, efficiency and power. The evaluation is simulation-based, permitting flexible application of the package to a broad range of scientific settings. Using lung cancer mortality data from Ohio, the package is illustrated with a detailed case-study in which two statistical goals are considered: (i) the evaluation of small-sample operating characteristics for two-phase and case-control designs and (ii) the planning and design of a future two-phase study.

Power and sample size for multivariate logistic modeling of unmatched case-control studies

“…Simulation code is also described in the Appendix. Alternative more general simulation methods were published previously(6). …”

“…The solid curve (“naïve”) is for a simulation study that adjusts for the confounder, education, but naively ignores any eventual need to control for non-differential recall error regarding benign breast biopsy status. The dashed curve (“realistic”) is for a simulation study that accounts for the confounder and for non-differential recall error using the method of Lyles and Lin(6, 11). …”

“…Simulations provide an alternative approach for sample size and power calculations for unmatched case-control data (6, 7). Not only can simulation allow for multivariate logistic modeling, it can also handle other potential complications such as measurement error.…”

Evaluating Public Health Interventions: 3. The Two-Stage Design for Confounding Bias Reduction—Having Your Cake and Eating It Two

“…In fact, packages for the analysis of two-stage designs are available for SAS users 23 and R users. 24 …”

Joint modeling of longitudinal and survival data with the Cox model and two-phase sampling

“…Here we favor the conditional score method, because many studies assess multiple biomarkers (e.g., vaccine efficacy trials) where the likelihood approach may computationally fail. However, to the best of our knowledge, no joint modeling approaches have focused on the common situation where the longitudinal biomarkers are measured on a designed sub-sample of the full study cohort, for example with a case-cohort sample (Prentice, 1986) or two-phase sample (Haneuse et al, 2011). We extend the full data conditional score method to handle general types of missing at random sub-sampling designs with Bernoulli sampling of subjects for measuring the biomarker trajectories, using inverse probability weighting (IPW) or augmented IPW (AIPW) to correct for biased sampling.…”