Minmax designs for planning the second phase in a two‐phase case–control study

Abstract

Two-phase designs, in which a subsample with validation or complete information data is sampled stratified both on outcome and covariate from a first-phase study with incomplete data, have been proposed over 10 years ago and have been proven to result in efficient estimates with respect to standard designs. The efficiency depends, however, on the sampling fractions within each stratum. Our aim is to present a strategy for obtaining an optimized design, i.e. sampling fractions, that makes use of available phase-one data of an existing case-control study considered as the first-phase sample when the focus is on estimating a parameter vector. No global optimal design exists and local optimal designs depend on scenarios comprising the true disease model and the association between the phase-one and phase-two information. We develop an admissibility test that rejects scenarios inconsistent with the phase-one data and, for the selected scenarios, determine a minmax D- or A-optimal design that protects against worst-case scenarios. This work is applied on two examples.

The Cost-Effectiveness of Reclassification Sampling for Prevalence Estimation

“…To resolve that problem, one can either reduce the number of parameters or add an equation to the system. Fujisawa and Izumi [10], as well as Sutcliffe [6], [7], introduce additional equations by requiring at least three classifications in order to estimate prevalence, sensitivity and specificity. It is possible, however, to reduce the number of parameters in the system with an alternative constraint and avoid a third classification.…”

“…The first option (one-phase sampling) involves classifying all individuals in the sample using the gold standard. However, when the gold standard is significantly more expensive than an imperfect classifier, two-phase sampling may be cost-effective [1]–[6]. Two-phase (or “double”) sampling involves classifying all individuals in the sample with the imperfect mechanism, and then reclassifying a subset of individuals in the sample with the gold standard.…”

Efficient sampling approaches to address confounding in database studies

“…We have recently extended this 45 to accommodate multiple confounding variables and/or covariates and either a binary or a categorical exposure; we also indicated how to proceed when exposure is represented as a quantitative variate. Schill and Wild 46 developed a strategy to obtain optimised sampling fractions to estimate a parameter vector. They note that no global optimal design exists and that local optimal designs depend on scenarios comprising the true disease model and the association between the phase-1 and phase-2 information.…”

A two‐stage phase II clinical trial design with nested criteria for early stopping and efficacy

“…We cite only some of the literature here. The original idea of the two‐stage design for phase II trials is attributed to R. Simon, with other developments building on the original idea . There are review articles covering phase II designs and, more generally, issues in clinical trials .…”

“…The original idea of the two-stage design for phase II trials is attributed to R. Simon,1 with other developments building on the original idea. [2][3][4][5][6][7][8] There are review articles covering phase II designs and, more generally, issues in clinical trials. [9][10][11][12][13] The present work is concerned with separate endpoints for futility and efficacy.…”