Variance identification and efficiency analysis in randomized experiments under the matched-pair design

In his 1923 landmark article, Neyman introduced randomization-based inference to estimate average treatment effects from experiments under the completely randomized design. Under this framework, Neyman considered the statistical estimation of the sample average treatment effect and derived the variance of the standard estimator using the treatment assignment mechanism as the sole basis of inference. In this paper, I extend Neyman's analysis to randomized experiments under the matched-pair design where experimental units are paired based on their pre-treatment characteristics and the randomization of treatment is subsequently conducted within each matched pair. I study the variance identification for the standard estimator of average treatment effects and analyze the relative efficiency of the matched-pair design over the completely randomized design. I also show how to empirically evaluate the relative efficiency of the two designs using experimental data obtained under the matched-pair design. My randomization-based analysis differs from previous studies in that it avoids modeling and other assumptions as much as possible. Finally, the analytical results are illustrated with numerical and empirical examples.     

An alternative empirical likelihood method in missing response problems and causal inference

Missing responses are common problems in medical, social, and economic studies. When responses are missing at random, a complete case data analysis may result in biases. A popular debias method is inverse probability weighting proposed by Horvitz and Thompson. To improve efficiency, Robins et al. proposed an augmented inverse probability weighting method. The augmented inverse probability weighting estimator has a double‐robustness property and achieves the semiparametric efficiency lower bound when the regression model and propensity score model are both correctly specified. In this paper, we introduce an empirical likelihood‐based estimator as an alternative to Qin and Zhang (2007). Our proposed estimator is also doubly robust and locally efficient. Simulation results show that the proposed estimator has better performance when the propensity score is correctly modeled. Moreover, the proposed method can be applied in the estimation of average treatment effect in observational causal inferences. Finally, we apply our method to an observational study of smoking, using data from the Cardiovascular Outcomes in Renal Atherosclerotic Lesions clinical trial. Copyright © 2016 John Wiley & Sons, Ltd.     

Analyzing medical costs with time‐dependent treatment: The nested g‐formula

As medical expenses continue to rise, methods to properly analyze cost outcomes are becoming of increasing relevance when seeking to compare average costs across treatments. Inverse probability weighted regression models have been developed to address the challenge of cost censoring in order to identify intent‐to‐treat effects (i.e., to compare mean costs between groups on the basis of their initial treatment assignment, irrespective of any subsequent changes to their treatment status). In this paper, we describe a nested g‐computation procedure that can be used to compare mean costs between two or more time‐varying treatment regimes. We highlight the relative advantages and limitations of this approach when compared with existing regression‐based models. We illustrate the utility of this approach as a means to inform public policy by applying it to a simulated data example motivated by costs associated with cancer treatments. Simulations confirm that inference regarding intent‐to‐treat effects versus the joint causal effects estimated by the nested g‐formula can lead to markedly different conclusions regarding differential costs. Therefore, it is essential to prespecify the desired target of inference when choosing between these two frameworks. The nested g‐formula should be considered as a useful, complementary tool to existing methods when analyzing cost outcomes.     

Models for the propensity score that contemplate the positivity assumption and their application to missing data and causality

Generalized linear models are often assumed to fit propensity scores, which are used to compute inverse probability weighted (IPW) estimators. To derive the asymptotic properties of IPW estimators, the propensity score is supposed to be bounded away from zero. This condition is known in the literature as strict positivity (or positivity assumption), and, in practice, when it does not hold, IPW estimators are very unstable and have a large variability. Although strict positivity is often assumed, it is not upheld when some of the covariates are unbounded. In real data sets, a data‐generating process that violates the positivity assumption may lead to wrong inference because of the inaccuracy in the estimations. In this work, we attempt to conciliate between the strict positivity condition and the theory of generalized linear models by incorporating an extra parameter, which results in an explicit lower bound for the propensity score. An additional parameter is added to fulfil the overlap assumption in the causal framework.     

A pilot design for observational studies: Using abundant data thoughtfully

Observational studies often benefit from an abundance of observational units. This can lead to studies that—while challenged by issues of internal validity—have inferences derived from sample sizes substantially larger than randomized controlled trials. But is the information provided by an observational unit best used in the analysis phase? We propose the use of a “pilot design,” in which observations are expended in the design phase of the study, and the posttreatment information from these observations is used to improve study design. In modern observational studies, which are data rich but control poor, pilot designs can be used to gain information about the structure of posttreatment variation. This information can then be used to improve instrumental variable designs, propensity score matching, doubly robust estimation, and other observational study designs. We illustrate one version of a pilot design, which aims to reduce within-set heterogeneity and improve performance in sensitivity analyses. This version of a pilot design expends observational units during the design phase to fit a prognostic model, avoiding concerns of overfitting. In addition, it enables the construction of “assignment-control plots,” which visualize the relationship between propensity and prognostic scores. We first show some examples of these plots, then we demonstrate in a simulation setting how this alternative use of the observations can lead to gains in terms of both treatment effect estimation and sensitivity analyses of unobserved confounding.     

A sequential Cox approach for estimating the causal effect of treatment in the presence of time‐dependent confounding applied to data from the Swiss HIV Cohort Study

When estimating the effect of treatment on HIV using data from observational studies, standard methods may produce biased estimates due to the presence of time‐dependent confounders. Such confounding can be present when a covariate, affected by past exposure, is both a predictor of the future exposure and the outcome. One example is the CD4 cell count, being a marker for disease progression for HIV patients, but also a marker for treatment initiation and influenced by treatment. Fitting a marginal structural model (MSM) using inverse probability weights is one way to give appropriate adjustment for this type of confounding. In this paper we study a simple and intuitive approach to estimate similar treatment effects, using observational data to mimic several randomized controlled trials. Each ‘trial’ is constructed based on individuals starting treatment in a certain time interval. An overall effect estimate for all such trials is found using composite likelihood inference. The method offers an alternative to the use of inverse probability of treatment weights, which is unstable in certain situations. The estimated parameter is not identical to the one of an MSM, it is conditioned on covariate values at the start of each mimicked trial. This allows the study of questions that are not that easily addressed fitting an MSM. The analysis can be performed as a stratified weighted Cox analysis on the joint data set of all the constructed trials, where each trial is one stratum. The model is applied to data from the Swiss HIV cohort study. Copyright © 2010 John Wiley & Sons, Ltd.     

Integrative modeling of multi‐platform genomic data under the framework of mediation analysis

Given the availability of genomic data, there have been emerging interests in integrating multi‐platform data. Here, we propose to model genetics (single nucleotide polymorphism (SNP)), epigenetics (DNA methylation), and gene expression data as a biological process to delineate phenotypic traits under the framework of causal mediation modeling. We propose a regression model for the joint effect of SNPs, methylation, gene expression, and their nonlinear interactions on the outcome and develop a variance component score test for any arbitrary set of regression coefficients. The test statistic under the null follows a mixture of chi‐square distributions, which can be approximated using a characteristic function inversion method or a perturbation procedure. We construct tests for candidate models determined by different combinations of SNPs, DNA methylation, gene expression, and interactions and further propose an omnibus test to accommodate different models. We then study three path‐specific effects: the direct effect of SNPs on the outcome, the effect mediated through expression, and the effect through methylation. We characterize correspondences between the three path‐specific effects and coefficients in the regression model, which are influenced by causal relations among SNPs, DNA methylation, and gene expression. We illustrate the utility of our method in two genomic studies and numerical simulation studies. Copyright © 2014 John Wiley & Sons, Ltd.     

Approaches to treatment effect heterogeneity in the presence of confounding

The literature on causal effect estimation tends to focus on the population mean estimand, which is less informative as medical treatments are becoming more personalized and there is increasing awareness that subpopulations of individuals may experience a group-specific effect that differs from the population average. In fact, it is possible that there is underlying systematic effect heterogeneity that is obscured by focusing on the population mean estimand. In this context, understanding which covariates contribute to this treatment effect heterogeneity (TEH) and how these covariates determine the differential treatment effect (TE) is an important consideration. Towards such an understanding, this paper briefly reviews three approaches used in making causal inferences and conducts a simulation study to compare these approaches according to their performance in an exploratory evaluation of TEH when the heterogeneous subgroups are not known a priori. Performance metrics include the detection of any heterogeneity, the identification and characterization of heterogeneous subgroups, and unconfounded estimation of the TE within subgroups. The methods are then deployed in a comparative effectiveness evaluation of drug-eluting versus bare-metal stents among 54 099 Medicare beneficiaries in the continental United States admitted to a hospital with acute myocardial infarction in 2008.     

Causal inference for community‐based multi‐layered intervention study

Estimating causal treatment effect for randomized controlled trials under post‐treatment confounding, that is, noncompliance and informative dropouts, is becoming an important problem in intervention/prevention studies when the treatment exposures are not completely controlled. When confounding is present in a study, the traditional intention‐to‐treat approach could underestimate the treatment effect because of insufficient exposure of treatment. In the recent two decades, many papers have been published to address such confounders to investigate the causal relationship between treatment and outcome of interest based on different modeling strategies. Most of the existing approaches, however, are suitable only for standard experiments. In this paper, we propose a new class of structural functional response model to address post‐treatment confounding in complex multi‐layered intervention studies within a longitudinal data setting. The new approach offers robust inference and is readily implemented. We illustrate and assess the performance of the proposed structural functional response model using both real and simulated data. Copyright © 2014 John Wiley & Sons, Ltd.     

A Bayesian sensitivity model for intention‐to‐treat analysis on binary outcomes with dropouts

Intention‐to‐treat (ITT) analysis is commonly used in randomized clinical trials. However, the use of ITT analysis presents a challenge: how to deal with subjects who drop out. Here we focus on randomized trials where the primary outcome is a binary endpoint. Several approaches are available for including the dropout subject in the ITT analysis, mainly chosen prior to unblinding the study. These approaches reduce the potential bias due to breaking the randomization code. However, the validity of the results will highly depend on untestable assumptions about the dropout mechanism. Thus, it is important to evaluate the sensitivity of the results across different missing‐data mechanisms. We propose here a Bayesian pattern‐mixture model for ITT analysis of binary outcomes with dropouts that applies over different types of missing‐data mechanisms. We introduce a new parameterization to identify the model, which is then used for sensitivity analysis. The parameterization is defined as the odds ratio of having an endpoint between the subjects who dropped out and those who completed the study. Such parameterization is intuitive and easy to use in sensitivity analysis; it also incorporates most of the available methods as special cases. The model is applied to TRial Of Preventing HYpertension. Copyright © 2008 John Wiley & Sons, Ltd.     

Penalized regression procedures for variable selection in the potential outcomes framework

A recent topic of much interest in causal inference is model selection. In this article, we describe a framework in which to consider penalized regression approaches to variable selection for causal effects. The framework leads to a simple ‘impute, then select’ class of procedures that is agnostic to the type of imputation algorithm as well as penalized regression used. It also clarifies how model selection involves a multivariate regression model for causal inference problems and that these methods can be applied for identifying subgroups in which treatment effects are homogeneous. Analogies and links with the literature on machine learning methods, missing data, and imputation are drawn. A difference least absolute shrinkage and selection operator algorithm is defined, along with its multiple imputation analogs. The procedures are illustrated using a well‐known right‐heart catheterization dataset. Copyright © 2015 John Wiley & Sons, Ltd.     

A causal model for longitudinal randomised trials with time‐dependent non‐compliance

In the presence of non‐compliance, conventional analysis by intention‐to‐treat provides an unbiased comparison of treatment policies but typically under‐estimates treatment efficacy. With all‐or‐nothing compliance, efficacy may be specified as the complier‐average causal effect (CACE), where compliers are those who receive intervention if and only if randomised to it. We extend the CACE approach to model longitudinal data with time‐dependent non‐compliance, focusing on the situation in which those randomised to control may receive treatment and allowing treatment effects to vary arbitrarily over time. Defining compliance type to be the time of surgical intervention if randomised to control, so that compliers are patients who would not have received treatment at all if they had been randomised to control, we construct a causal model for the multivariate outcome conditional on compliance type and randomised arm. This model is applied to the trial of alternative regimens for glue ear treatment evaluating surgical interventions in childhood ear disease, where outcomes are measured over five time points, and receipt of surgical intervention in the control arm may occur at any time. We fit the models using Markov chain Monte Carlo methods to obtain estimates of the CACE at successive times after receiving the intervention. In this trial, over a half of those randomised to control eventually receive intervention. We find that surgery is more beneficial than control at 6months, with a small but non‐significant beneficial effect at 12months. © 2015 The Authors. Statistics in Medicine Published by JohnWiley & Sons Ltd.     

Allowing for uncertainty due to missing continuous outcome data in pairwise and network meta‐analysis

Missing outcome data are commonly encountered in randomized controlled trials and hence may need to be addressed in a meta‐analysis of multiple trials. A common and simple approach to deal with missing data is to restrict analysis to individuals for whom the outcome was obtained (complete case analysis). However, estimated treatment effects from complete case analyses are potentially biased if informative missing data are ignored. We develop methods for estimating meta‐analytic summary treatment effects for continuous outcomes in the presence of missing data for some of the individuals within the trials. We build on a method previously developed for binary outcomes, which quantifies the degree of departure from a missing at random assumption via the informative missingness odds ratio. Our new model quantifies the degree of departure from missing at random using either an informative missingness difference of means or an informative missingness ratio of means, both of which relate the mean value of the missing outcome data to that of the observed data. We propose estimating the treatment effects, adjusted for informative missingness, and their standard errors by a Taylor series approximation and by a Monte Carlo method. We apply the methodology to examples of both pairwise and network meta‐analysis with multi‐arm trials. © 2014 The Authors. Statistics in Medicine Published by John Wiley & Sons Ltd.     

Targeted estimation and inference for the sample average treatment effect in trials with and without pair‐matching

In cluster randomized trials, the study units usually are not a simple random sample from some clearly defined target population. Instead, the target population tends to be hypothetical or ill‐defined, and the selection of study units tends to be systematic, driven by logistical and practical considerations. As a result, the population average treatment effect (PATE) may be neither well defined nor easily interpretable. In contrast, the sample average treatment effect (SATE) is the mean difference in the counterfactual outcomes for the study units. The sample parameter is easily interpretable and arguably the most relevant when the study units are not sampled from some specific super‐population of interest. Furthermore, in most settings, the sample parameter will be estimated more efficiently than the population parameter. To the best of our knowledge, this is the first paper to propose using targeted maximum likelihood estimation (TMLE) for estimation and inference of the sample effect in trials with and without pair‐matching. We study the asymptotic and finite sample properties of the TMLE for the sample effect and provide a conservative variance estimator. Finite sample simulations illustrate the potential gains in precision and power from selecting the sample effect as the target of inference. This work is motivated by the Sustainable East Africa Research in Community Health (SEARCH) study, a pair‐matched, community randomized trial to estimate the effect of population‐based HIV testing and streamlined ART on the 5‐year cumulative HIV incidence (NCT01864603). The proposed methodology will be used in the primary analysis for the SEARCH trial. Copyright © 2016 John Wiley & Sons, Ltd.     

A synthetic estimator for the efficacy of clinical trials with all‐or‐nothing compliance

A critical issue in the analysis of clinical trials is patients' noncompliance to assigned treatments. In the context of a binary treatment with all or nothing compliance, the intent‐to‐treat analysis is a straightforward approach to estimating the effectiveness of the trial. In contrast, there exist 3 commonly used estimators with varying statistical properties for the efficacy of the trial, formally known as the complier‐average causal effect. The instrumental variable estimator may be unbiased but can be extremely variable in many settings. The as treated and per protocol estimators are usually more efficient than the instrumental variable estimator, but they may suffer from selection bias. We propose a synthetic approach that incorporates all 3 estimators in a data‐driven manner. The synthetic estimator is a linear convex combination of the instrumental variable, per protocol, and as treated estimators, resembling the popular model‐averaging approach in the statistical literature. However, our synthetic approach is nonparametric; thus, it is applicable to a variety of outcome types without specific distributional assumptions. We also discuss the construction of the synthetic estimator using an analytic form derived from a simple normal mixture distribution. We apply the synthetic approach to a clinical trial for post‐traumatic stress disorder.     

Mendelian randomization analysis of a time‐varying exposure for binary disease outcomes using functional data analysis methods

A Mendelian randomization (MR) analysis is performed to analyze the causal effect of an exposure variable on a disease outcome in observational studies, by using genetic variants that affect the disease outcome only through the exposure variable. This method has recently gained popularity among epidemiologists given the success of genetic association studies. Many exposure variables of interest in epidemiological studies are time varying, for example, body mass index (BMI). Although longitudinal data have been collected in many cohort studies, current MR studies only use one measurement of a time‐varying exposure variable, which cannot adequately capture the long‐term time‐varying information. We propose using the functional principal component analysis method to recover the underlying individual trajectory of the time‐varying exposure from the sparsely and irregularly observed longitudinal data, and then conduct MR analysis using the recovered curves. We further propose two MR analysis methods. The first assumes a cumulative effect of the time‐varying exposure variable on the disease risk, while the second assumes a time‐varying genetic effect and employs functional regression models. We focus on statistical testing for a causal effect. Our simulation studies mimicking the real data show that the proposed functional data analysis based methods incorporating longitudinal data have substantial power gains compared to standard MR analysis using only one measurement. We used the Framingham Heart Study data to demonstrate the promising performance of the new methods as well as inconsistent results produced by the standard MR analysis that relies on a single measurement of the exposure at some arbitrary time point.     

Martingale residual‐based method to control for confounders measured only in a validation sample in time‐to‐event analysis

Unmeasured confounding remains an important problem in observational studies, including pharmacoepidemiological studies of large administrative databases. Several recently developed methods utilize smaller validation samples, with information on additional confounders, to control for confounders unmeasured in the main, larger database. However, up‐to‐date applications of these methods to survival analyses seem to be limited to propensity score calibration, which relies on a strong surrogacy assumption. We propose a new method, specifically designed for time‐to‐event analyses, which uses martingale residuals, in addition to measured covariates, to enhance imputation of the unmeasured confounders in the main database. The method is applicable for analyses with both time‐invariant data and time‐varying exposure/confounders. In simulations, our method consistently eliminated bias because of unmeasured confounding, regardless of surrogacy violation and other relevant design parameters, and almost always yielded lower mean squared errors than other methods applicable for survival analyses, outperforming propensity score calibration in several scenarios. We apply the method to a real‐life pharmacoepidemiological database study of the association between glucocorticoid therapy and risk of type II diabetes mellitus in patients with rheumatoid arthritis, with additional potential confounders available in an external validation sample. Compared with conventional analyses, which adjust only for confounders measured in the main database, our estimates suggest a considerably weaker association. Copyright © 2016 John Wiley & Sons, Ltd.     

Repairing the efficiency loss due to varying cluster sizes in two‐level two‐armed randomized trials with heterogeneous clustering

In two‐armed trials with clustered observations the arms may differ in terms of (i) the intraclass correlation, (ii) the outcome variance, (iii) the average cluster size, and (iv) the number of clusters. For a linear mixed model analysis of the treatment effect, this paper examines the expected efficiency loss due to varying cluster sizes based upon the asymptotic relative efficiency of varying versus constant cluster sizes. Simple, but nearly cost‐optimal, correction factors are derived for the numbers of clusters to repair this efficiency loss. In an extensive Monte Carlo simulation, the accuracy of the asymptotic relative efficiency and its Taylor approximation are examined for small sample sizes. Practical guidelines are derived to correct the numbers of clusters calculated under constant cluster sizes (within each treatment) when planning a study. Because of the variety of simulation conditions, these guidelines can be considered conservative but safe in many realistic situations. Copyright © 2016 John Wiley & Sons, Ltd.