On the analysis of two‐phase designs in cluster‐correlated data settings

Abstract

In public health research, information that is readily available may be insufficient to address the primary question(s) of interest. One cost‐efficient way forward, especially in resource‐limited settings, is to conduct a two‐phase study in which the population is initially stratified, at phase I, by the outcome and/or some categorical risk factor(s). At phase II detailed covariate data is ascertained on a subsample within each phase I strata. While analysis methods for two‐phase designs are well established, they have focused exclusively on settings in which participants are assumed to be independent. As such, when participants are naturally clustered (eg, patients within clinics) these methods may yield invalid inference. To address this, we develop a novel analysis approach based on inverse‐probability weighting that permits researchers to specify some working covariance structure and appropriately accounts for the sampling design and ensures valid inference via a robust sandwich estimator for which a closed‐form expression is provided. To enhance statistical efficiency, we propose a calibrated inverse‐probability weighting estimator that makes use of information available at phase I but not used in the design. In addition to describing the technique, practical guidance is provided for the cluster‐correlated data settings that we consider. A comprehensive simulation study is conducted to evaluate small‐sample operating characteristics, including the impact of using naïve methods that ignore correlation due to clustering, as well as to investigate design considerations. Finally, the methods are illustrated using data from a one‐time survey of the national antiretroviral treatment program in Malawi.

Sampling strategies to evaluate the prognostic value of a new biomarker on a time-to-event end-point

“…The sample size of retrospective studies is often planned considering budget constrains rather than a proper evaluation of the statistical power [ 10 ], also due to the lack of methodologies for power calculation in this setting. Cai and Zeng [ 11 ] focused on power in case–cohort design without any stratification; Haneuse et al [ 12 , 13 ] focused on binary outcomes, but a general strategy for power evaluation is missing for survival data.…”

Using Big Data to Estimate Dementia Prevalence in New Zealand: Protocol for an Observational Study

“…The choice of a loss function is usually a balance between the goal of the analysis and the efficiency and complexity of the function. GEE is a well-known method for regression in the presence of correlated data or repeated measures [ 15 , 16 ]. The efficiency of GEE depends on the assumptions made about the variability of the data.…”

Small‐sample inference for cluster‐based outcome‐dependent sampling schemes in resource‐limited settings: Investigating low birthweight in Rwanda

“…Examples of outcome-dependent sampling designs include the case-control, nested case-control, casecohort, and two-phase designs (Prentice and Pyke, 1979;Breslow and Day, 1980;Langholz and Thomas, 1990;Wacholder, 1991;White, 1982). For the most part, the literature on ODS has focused on the setting where individuals are treated as independent, although methods have been recently proposed for longitudinal and cluster-correlated data settings (Neuhaus et al, 2002(Neuhaus et al, , 2006Schildcrout and Rathouz, 2010;Schildcrout et al, 2013;Neuhaus et al, 2014;Haneuse and Rivera-Rodriguez, 2018;Rivera-Rodriguez et al, 2019). However, in Biometrics, 000 0000 both the independent and correlated data settings, the majority of designs proposed involve sampling at the level of the individual (i.e.…”