Pooling Designs for Outcomes under a Gaussian Random Effects Model

Abstract

Summary Due to the rising cost of laboratory assays, it has become increasingly common in epidemiological studies to pool biospecimens. This is particularly true in longitudinal studies, where the cost of performing multiple assays over time can be prohibitive. In this article, we consider the problem of estimating the parameters of a Gaussian random effects model when the repeated outcome is subject to pooling. We consider different pooling designs for the efficient maximum likelihood estimation of variance components, with particular attention to estimating the intraclass correlation coefficient. We evaluate the efficiencies of different pooling design strategies using analytic and simulation study results. We examine the robustness of the designs to skewed distributions and consider unbalanced designs. The design methodology is illustrated with a longitudinal study of premenopausal women focusing on assessing the reproducibility of F2-isoprostane, a biomarker of oxidative stress, over the menstrual cycle.

Estimating covariate‐adjusted measures of diagnostic accuracy based on pooled biomarker assessments

“…Regretfully, there does not exist formal or informal techniques that can be used for assessing the validity of this assumption, when one has access to pooled data only. Consequently, the development of such techniques could be a beneficial topic for future research, especially since this assumption is common among the associated literature (e.g., see Faraggi et al, 2003; Liu and Schisterman, 2003; Mumford et al, 2006; Bondell et al, 2007; Vexler et al, 2008; Malinovsky et al, 2012). …”

“…The key feature of pooling that facilitates a reduction in testing cost is that instead of measuring biomarker specimens one by one, pools (of multiple specimens) are constructed and measured. Biomarker assessments based on pooled samples result in a continuous value, which is typically assumed to represent the average (or sum) of the biomarker concentrations over the pooled individuals (e.g., see Faraggi et al, 2003; Liu and Schisterman, 2003; Schisterman et al, 2004; Mumford et al, 2006; Bondell et al, 2007; Vexler et al, 2008; Malinovsky et al, 2012). The use of pooling can offer a drastic reduction in data collection costs.…”

“…Statistical research in biomarker pooling has predominantly focused on developing methods that can estimate the unadjusted receiver-operating characteristic (ROC) curve and corresponding area under the curve (AUC) through the analysis of pooled biomarker assessments (Faraggi et al, 2003; Liu and Schisterman, 2003; Mumford et al, 2006; Bondell et al, 2007; Vexler et al, 2008; Malinovsky et al, 2012). The ROC curve and AUC are fundamental measures that summarize the discriminatory ability of a biomarker of interest, with respect to correctly distinguishing between cases (diseased) and controls (nondiseased).…”

Regression analysis for multiple‐disease group testing data

“…This would involve simply letting the j subscript in our notation keep track of the time points for the i th individual in the k th group. One potential limitation with this extension is that the same individuals would need to be in the same groups at each time point, although this design has been proposed in related problems where pooling is used [33]. …”

Case‐control data analysis for randomly pooled biomarkers

“…Pooling is commonly used as a cost-saving tool in screening for disease (e.g., screening donated blood for HIV), but also has utility when continuous biomarkers are assessed for epidemiological studies (Kacena et al, 1998;Pilcher et al, 2005). Strategic pooling designs have been shown to promote statistical efficiency, particularly when compared with similar designs that select the same number of individual assays for analysis (Weinberg and Umbach, 1999;Liu and Schisterman, 2003;Mumford et al, 2006;Ma et al, 2011;Schisterman et al, 2011;Malinovsky et al, 2012;Saha-Chaudhuri and Weinberg, 2013;Heffernan et al, 2014). In certain situations assays may require a minimum volume for analysis that is impractical or impossible to obtain from single specimens.…”