Bayesian variable selection for high dimensional predictors and self-reported outcomes.

BACKGROUND: The onset of silent diseases such as type 2 diabetes is often registered through self-report in large prospective cohorts. Self-reported outcomes are cost-effective; however, they are subject to error. Diagnosis of silent events may also occur through the use of imperfect laboratory-based diagnostic tests. In this paper, we describe an approach for variable selection in high dimensional datasets for settings in which the outcome is observed with error. METHODS: We adapt the spike and slab Bayesian Variable Selection approach in the context of error-prone, self-reported outcomes. The performance of the proposed approach is studied through simulation studies. An illustrative application is included using data from the Women's Health Initiative SNP Health Association Resource, which includes extensive genotypic (>900,000 SNPs) and phenotypic data on 9,873 African American and Hispanic American women. RESULTS: Simulation studies show improved sensitivity of our proposed method when compared to a naive approach that ignores error in the self-reported outcomes. Application of the proposed method resulted in discovery of several single nucleotide polymorphisms (SNPs) that are associated with risk of type 2 diabetes in a dataset of 9,873 African American and Hispanic participants in the Women's Health Initiative. There was little overlap among the top ranking SNPs associated with type 2 diabetes risk between the racial groups, adding support to previous observations in the literature of disease associated genetic loci that are often not generalizable across race/ethnicity populations. The adapted Bayesian variable selection algorithm is implemented in R. The source code for the simulations are available in the Supplement. CONCLUSIONS: Variable selection accuracy is reduced when the outcome is ascertained by error-prone self-reports. For this setting, our proposed algorithm has improved variable selection performance when compared to approaches that neglect to account for the error-prone nature of self-reports.

A Modified Random Survival Forests Algorithm for High Dimensional Predictors and Self-Reported Outcomes.

We present an ensemble tree-based algorithm for variable selection in high dimensional datasets, in settings where a time-to-event outcome is observed with error. The proposed methods are motivated by self-reported outcomes collected in large-scale epidemiologic studies, such as the Women's Health Initiative. The proposed methods equally apply to imperfect outcomes that arise in other settings such as data extracted from electronic medical records. To evaluate the performance of our proposed algorithm, we present results from simulation studies, considering both continuous and categorical covariates. We illustrate this approach to discover single nucleotide polymorphisms that are associated with incident Type II diabetes in the Women's Health Initiative. A freely available R package icRSF (R Core Team, 2018; Xu et al., 2018) has been developed to implement the proposed methods.

Study design for non-recurring, time-to-event outcomes in the presence of error-prone diagnostic tests or self-reports.

Sequentially administered, laboratory-based diagnostic tests or self-reported questionnaires are often used to determine the occurrence of a silent event. In this paper, we consider issues relevant in design of studies aimed at estimating the association of one or more covariates with a non-recurring, time-to-event outcome that is observed using a repeatedly administered, error-prone diagnostic procedure. The problem is motivated by the Women's Health Initiative, in which diabetes incidence among the approximately 160,000 women is obtained from annually collected self-reported data. For settings of imperfect diagnostic tests or self-reports with known sensitivity and specificity, we evaluate the effects of various factors on resulting power and sample size calculations and compare the relative efficiency of different study designs. The methods illustrated in this paper are readily implemented using our freely available R software package icensmis, which is available at the Comprehensive R Archive Network website. An important special case is that when diagnostic procedures are perfect, they result in interval-censored, time-to-event outcomes. The proposed methods are applicable for the design of studies in which a time-to-event outcome is interval censored. Copyright © 2016 John Wiley & Sons, Ltd.

Marginal structural models for the estimation of the risk of Diabetes Mellitus in the presence of elevated depressive symptoms and antidepressant medication use in the Women's Health Initiative observational and clinical trial cohorts.

BACKGROUND: We evaluate the combined effect of the presence of elevated depressive symptoms and antidepressant medication use with respect to risk of type 2 diabetes among approximately 120,000 women enrolled in the Women's Health Initiative (WHI), and compare several different statistical models appropriate for causal inference in non-randomized settings. METHODS: Data were analyzed for 52,326 women in the Women's Health Initiative Clinical Trials (CT) Cohort and 68,169 women in the Observational Study (OS) Cohort after exclusions. We included follow-up to 2005, resulting in a median duration of 7.6 years of follow up after enrollment. Results from three multivariable Cox models were compared to those from marginal structural models that included time varying measures of antidepressant medication use, presence of elevated depressive symptoms and BMI, while adjusting for potential confounders including age, ethnicity, education, minutes of recreational physical activity per week, total energy intake, hormone therapy use, family history of diabetes and smoking status. RESULTS: Our results are consistent with previous studies examining the relationship of antidepressant medication use and risk of type 2 diabetes. All models showed a significant increase in diabetes risk for those taking antidepressants. The Cox Proportional Hazards models using baseline covariates showed the lowest increase in risk , with hazard ratios of 1.19 (95 % CI 1.06 - 1.35) and 1.14 (95 % CI 1.01 - 1.30) in the OS and CT, respectively. Hazard ratios from marginal structural models comparing antidepressant users to non-users were 1.35 (95 % CI 1.21 - 1.51) and 1.27 (95 % CI 1.13 - 1.43) in the WHI OS and CT, respectively - however, differences among estimates from traditional Cox models and marginal structural models were not statistically significant in both cohorts. One explanation suggests that time-dependent confounding was not a substantial factor in these data, however other explanations exist. Unadjusted Cox Proportional Hazards models showed that women with elevated depressive symptoms had a significant increase in diabetes risk that remained after adjustment for confounders. However, this association missed the threshold for statistical significance in propensity score adjusted and marginal structural models. CONCLUSIONS: Results from the multiple approaches provide further evidence of an increase in risk of type 2 diabetes for those on antidepressants.

SEMIPARAMETRIC TIME TO EVENT MODELS IN THE PRESENCE OF ERROR-PRONE, SELF-REPORTED OUTCOMES-WITH APPLICATION TO THE WOMEN'S HEALTH INITIATIVE.

The onset of several silent, chronic diseases such as diabetes can be detected only through diagnostic tests. Due to cost considerations, self-reported outcomes are routinely collected in lieu of expensive diagnostic tests in large-scale prospective investigations such as the Women's Health Initiative. However, self-reported outcomes are subject to imperfect sensitivity and specificity. Using a semiparametric likelihood-based approach, we present time to event models to estimate the association of one or more covariates with a error-prone, self-reported outcome. We present simulation studies to assess the effect of error in self-reported outcomes with regard to bias in the estimation of the regression parameter of interest. We apply the proposed methods to prospective data from 152,830 women enrolled in the Women's Health Initiative to evaluate the effect of statin use with the risk of incident diabetes mellitus among postmenopausal women. The current analysis is based on follow-up through 2010, with a median duration of follow-up of 12.1 years. The methods proposed in this paper are readily implemented using our freely available R software package icensmis, which is available at the Comprehensive R Archive Network (CRAN) website.

Stratified Weibull Regression Model for Interval-Censored Data.

Interval censored outcomes arise when a silent event of interest is known to have occurred within a specific time period determined by the times of the last negative and first positive diagnostic tests. There is a rich literature on parametric and non-parametric approaches for the analysis of interval-censored outcomes. A commonly used strategy is to use a proportional hazards (PH) model with the baseline hazard function parameterized. The proportional hazards assumption can be relaxed in stratified models by allowing the baseline hazard function to vary across strata defined by a subset of explanatory variables. In this paper, we describe and implement a new R package straweib, for fitting a stratified Weibull model appropriate for interval censored outcomes. We illustrate the R package straweib by analyzing data from a longitudinal oral health study on the timing of the emergence of permanent teeth in 4430 children.