Search

This dissertation develops modern statistical methods, targeted maximum likelihood estimation (TMLE) and ensemble machine learning, for two common problems in epidemiology and public health: 1) estimating the population size based on capture recapture designs, and 2) estimating the continuous and discrete treatment effect of multiple exposures. For the first problem, We proposed novel target parameters for each identification assumption and robust estimators based on TMLE, provided the efficient influence curves for each the parameter, proved the statistical properties of the estimators, and applied them to data collected from national-level infectious disease surveillance systems. We also used simulations with identification assumption violations to test the reliability of the estimation. For the second problem, we developed our target parameter and TMLE estimators and applied them to various types of empirical data collected by community-level follow-up surveys and state-level electronic health record data. We provided simulations and sensitivity analysis to show the performance of the TMLE estimators compared to existing ones. In chapter 1, we gave an introduction to TMLE, the road map of targeted learning, and a more detailed summary of the following chapters. In chapter 2, we developed a novel approach to estimate the population size based on capture recapture designs and evaluated the estimation reliability. In chapter 3, we utilized the targeted learning approach to assess the performance of a diabetes care program on glycemic control of type 2 diabetes patients, and identified patient subgroups with most successful treatment effects. In chapter 4, we proposed a robust variable importance measure based on TMLE and applied it in estimating the transmission effects of mother’s eating behavior on the next generation.

In chapter 1 we study explained variation under the additive hazards regression model for right-censored data. We consider different approaches for developing such a measure, and focus on one that estimates the proportion of variation in the failure time explained by the covariates. We study the properties of the measure both analytically, and through extensive simulations. We apply the measure to a well-known survival dataset as well as the linked surveillance, epidemiology, and end results-Medicare database for prediction of mortality in early stage prostate cancer patients using high-dimensional claims codes.In chapter 2 we propose a new flexible method for survival prediction: DeepHazard, a neural network for time-varying risks. Prognostic models in survival analysis are aimed at understanding the relationship between patients’ covariates and the distribution of survival time. Traditionally, semiparametric models, such as the Cox model, have been assumed. These often rely on strong proportionality assumptions of the hazard that might be violated in practice. Moreover, they do not often include covariates' information updated over time. Our approach is tailored for a wide range of continuous hazards forms, with the only restriction of being additive in time. A flexible implementation, allowing different optimization methods, along with any norm penalty, is developed. Numerical examples illustrate that our approach outperforms existing state-of-the-art methodology in terms of predictive capability evaluated through the C-index metric. The same is revealed on the popular real datasets as METABRIC, GBSG, ACTG and PBC.In chapter 3 we consider the conditional treatment effect for competing risks data in observational studies. While it is described as a constant difference between the hazard functions given the covariates, we do not assume the additive hazards model in order to adjust for the covariates. We derive the efficient score for the treatment effect using modern semiparametric theory, as well as two doubly robust scores with respect to both the assumed propensity score for treatment and the censoring model, and the outcome models for the competing risks. We provide the asymptotic distributions of the estimators when the two sets of working models are both correct, or when only one of them is correct. We study the inference based on these estimators using simulation. The estimators are applied to the data from a cohort of Japanese men in Hawaii followed since 1960s in order to study the effect of midlife drinking behavior on late life cognitive outcomes. In chapter 4 we consider doubly robust estimation of the causal hazard ratio in observational studies. The treatment effect of interest, described as the constant ratio between the hazard functions of thetwo potential outcomes, is parametrized by the Marginal Structural Cox Model. Under the assumption of no unmeasured confounders, causal methods, as Cox-IPW, have been developed for estimation of the treatment effect of interest. However no doubly robust methods have been proposed under the Marginal Structural Cox model. We develop an AIPW estimator for this popular model that is both model and rate-doubly robust with respect to the treatment assignment modeland the conditional outcome model. The proposed estimator is applied to the data from a cohort of Japanese men in Hawaii followed since 1960s in order to study the effect of mid-life alcohol exposure on overall death.

Considerable thought is often put into designing randomized control trials (RCTs). From power analyses and complex sampling designs implemented preintervention to nuanced quasi-experimental models used to estimate treatment effects postintervention, RCT design can be quite complicated. Yet when psychological constructs measured using survey scales are the outcome of interest, measurement is often an afterthought, even in RCTs. The purpose of this study is to examine how choices about scoring and calibration of survey item responses affect recovery of true treatment effects. Specifically, simulation and empirical studies are used to compare the performance of sum scores, which are frequently used in RCTs in psychology and education, to that of approaches rooted in item response theory (IRT) that better account for the longitudinal, multigroup nature of the data. The results from this study indicate that selecting an IRT model that matches the nature of the data can significantly reduce bias in treatment effect estimates and reduce standard errors.

A new causal inference method called the Stability-Controlled Quasi-Experiment (SCQE) was recently proposed for identifying the effect of a treatment on units in a population that are treated. It allows for non-randomized or otherwise unknown means of selection into treatment, relying only on an assumption about the stability of the non-treatment outcomes of the population. We apply SCQE to the case of a national public health intervention in Tanzania to prevent active tuberculosis in HIV-positive patients by prescribing Isoniazid (IPT). Our preliminary results suggest much of the treatment is given to patients at low risk for developing active tuberculosis. In the process, we offer best practices and discuss challenges faced during its implementation. We share steps currently underway to generate confidence in these results, extend the method further, and derive valid standard errors.

In this paper, a propensity score-integrated power prior approach is developed to augment the control arm of a two-arm randomized controlled trial (RCT) with subjects from multiple external data sources such as real-world data (RWD) and historical clinical studies containing subject-level outcomes and covariates. The propensity scores for the subjects in the external data sources versus the subjects in the RCT are first estimated, and then subjects are placed in different strata based on their estimated propensity scores. Within each propensity score stratum, a power prior is formulated with the information contributed by the external data sources, and Bayesian inference on the treatment effect is obtained. The proposed approach is implemented under the two-stage study design framework utilizing the outcome-free principle to ensure the integrity of a study. An illustrative example is provided to demonstrate the implementation of the proposed approach.

A common assumption of data analysis in clinical trials is that the patient population, as well as treatment effects, do not vary during the course of the study. However, when trials enroll patients over several years, this hypothesis may be violated. Ignoring variations of the outcome distributions over time, under the control and experimental treatments, can lead to biased treatment effect estimates and poor control of false positive results. We propose and compare two procedures that account for possible variations of the outcome distributions over time, to correct treatment effect estimates, and to control type-I error rates. The first procedure models trends of patient outcomes with splines. The second leverages conditional inference principles, which have been introduced to analyze randomized trials when patient prognostic profiles are unbalanced across arms. These two procedures are applicable in response-adaptive clinical trials. We illustrate the consequences of trends in the outcome distributions in response-adaptive designs and in platform trials, and investigate the proposed methods in the analysis of a glioblastoma study.

In cluster-randomized trials, sometimes the effect of the intervention being studied differs between clusters, commonly referred to as treatment effect heterogeneity. In the analysis of stepped wedge and cluster-randomized crossover trials, it is possible to include terms in outcome regression models to allow for such treatment effect heterogeneity yet this is not frequently considered. Outside of some simulation studies of specific cases where the outcome is binary, the impact of failing to include terms for treatment effect heterogeneity on the variance of the treatment effect estimator is unknown. We analytically examine the impact of failing to include terms for treatment effect heterogeneity on the variance of the treatment effect estimator, when outcomes are continuous. Using analysis of variance and feasible generalized least squares we provide expressions for this variance. For both the cluster-randomized crossover design and the stepped wedge design, our analytic derivations indicate that failing to include treatment effect heterogeneity results in the estimates for variance of the treatment effect that are too small, leading to inflation of type I error rates. We therefore recommend assessing the sensitivity of sample size calculations and conclusions drawn from the analysis of cluster randomized trials to the inclusion of treatment effect heterogeneity.

To assess the effect of treatment in survival studies, a randomized controlled trial (RCTs) is usually designed. Due to disease progression or other safety reasons, patients in the standard group m...

This paper develops an empirical balancing approach for the estimation of treatment effects under two-sided noncompliance using a binary conditionally independent instrumental variable. The method weighs both treatment and outcome information with inverse probabilities to produce exact finite sample balance across instrument level groups. It is free of functional form assumptions on the outcome or the treatment selection step. By tailoring the loss function for the instrument propensity scores, the resulting treatment effect estimates exhibit both low bias and a reduced variance in finite samples compared to conventional inverse probability weighting methods. The estimator is automatically weight normalized and has similar bias properties compared to conventional two-stage least squares estimation under constant causal effects for the compliers. We provide conditions for asymptotic normality and semiparametric efficiency and demonstrate how to utilize additional information about the treatment selection step for bias reduction in finite samples. The method can be easily combined with regularization or other statistical learning approaches to deal with a high-dimensional number of observed confounding variables. Monte Carlo simulations suggest that the theoretical advantages translate well to finite samples. The method is illustrated in an empirical example.

The regression discontinuity design (RDD) is one of the most credible methods for causal inference that is often regarded as a missing data problem in the potential outcomes framework. However, the...

This article introduces a crossover design that is often used in clinical studies, with the advantage of comparing treatment effects within one study subject. In particular, the advantages and disadvantages of the two-period, two-sequence crossover design (2 × 2 or AB/BA crossover design), which is widely used in clinical practice, are identified, and the elements necessary for analysis are introduced. This article explains the carryover effect, period effect, sequence effect, and period-by-treatment interaction in a crossover design and examines the analysis commands of SAS along with example data. After confirming the carryover effect using a general linear model, the treatment effect is analyzed using a linear mixed effect model.

Statistical analyses of randomized controlled trials (RCTs) yield a causally valid estimate of the overall treatment effect, which is the contrast between the outcomes in two randomized treatment groups commonly accompanied by a confidence interval. In addition, the trial investigators may want to examine whether the observed treatment effect varies across patient subgroups (also called 'heterogeneity of treatment effects'), i.e. whether the treatment effect is modified by the value of a variable assessed at baseline. The statistical approach for this evaluation of potential effect modifiers is a test for statistical interaction to evaluate whether the treatment effect varies across levels of the effect modifier. In this article, we provide a concise and nontechnical explanation of the use of simple statistical tests for interaction to identify effect modifiers in RCTs. We explain how to calculate the test of interaction by hand, applied to a dataset with simulated data on 1,000 imaginary participants for illustration. (C) 2021 The Author(s). Published by Elsevier Inc.

In many applications such as case-control studies with contaminated controls, or the test of a treatment effect in the presence of nonresponders in biological experiments or clinical trials, a two-sample problem with one of the samples having a mixture structure often arises. Due to the importance and wide applications of scale mixtures and location mixtures, we consider in this paper the case that the component densities differ only in scale parameters and the case that the component densities differ only in location parameters, and further construct an EM-test for the two-sample problem under each case. We show that both the EM-tests possess a chi-squared null limiting distribution. The local power analysis and sample size calculations are also investigated. Finally, the simulation studies and real data analysis demonstrate that the proposed EM-tests have better performance than the existing methods.

Introduction Quantitative patient-reported outcome (PRO) measures ideally are analyzed on their original scales and responder analyses are used to aid the interpretation of those primary analyses. As stated in the FDA PRO Guidance for Medical Product Development (2009), one way to lend meaning and interpretation to such a PRO measure is to dichotomize between values where within-patient changes are considered clinically important and those that are not. But even a PRO scale with a cutoff score that discriminates well between responder and non-responders is fraught with some misclassification. Methods Using estimates of sensitivity and specificity on classification of responder status from a PRO instrument, formulas are provided to correct for such responder misclassification under the assumption of no treatment misclassification. Two case studies from sexual medicine illustrate the methodology. Results Adjustment formulas on cell counts for responder misclassification are a direct extension of correction formulas for misclassification on disease from a two-way cross-classification table of disease (yes, no) and exposure (yes, no). Unadjusted and adjusted estimates of treatment effect are compared in terms of odds ratio, response ratio, and response difference. In the two case studies, there was considerable underestimation of treatment effect. Discussion and conclusions The methodology can be applied to different therapeutic areas. Limitations of the methodology, such as when adjusted cell estimates become negative, are highlighted. The role of anchor-based methodology is discussed for obtaining estimates of sensitivity and specificity on responder classification. Correction for treatment effect bias from misclassification of responder status on PRO measures can lead to more trustworthy interpretation and effective decision-making. Clinicaltrials.gov: NCT00343200

Summary Background At the clinical trial design stage, assumptions regarding the treatment effects to be detected should be appropriate so that the required sample size can be calculated. There is evidence in the medical literature that sample size assumption can be overoptimistic. The aim of this study was to compare the distribution of the assumed effects versus that of the observed effects as a proxy for overoptimistic treatment effect assumptions at the study design stage. Materials and method Systematic reviews (SRs) published between 1 January 2010 and 31 December 2019 containing at least one meta-analysis on continuous outcomes were identified electronically. SR and primary study level characteristics were extracted from the SRs and the individual trials. Details on the sample size calculation process and assumptions and the observed treatment effects were extracted. Results Eighty-five SRs with meta-analysis containing 347 primary trials were included. The median number of SR authors was 5 (interquartile range: 4–7). At the primary study level, the majority were single centre (78.1%), utilized a parallel design (52%), and rated as an unclear/moderate level of risk of bias (34.3%). A sample size was described in only 31.7% (110/347) of studies. From this cohort of 110 studies, in only 37 studies was the assumed clinical difference that the study was designed to detect reported (37/110). The assumed treatment effect was recalculated for the remaining 73 studies (73/110). The one-sided exact signed rank test showed a significant difference between the assumed and observed treatment effects (P < 0.001) suggesting greater values for the assumed effect sizes. Conclusions Careful consideration of the assumptions at the design stage of orthodontic studies are necessary in order to reduce the unreliability of clinical study results and research waste.

ObjectiveTo develop an index to evaluate the bias attributable to composite outcomes (BACO) in comparative clinical studies.Study design and SettingI defined the BACO index as the ratio of the logarithm of the association measure (e.g., relative risk) of the composite outcome to that of its most relevant component endpoint (e.g., mortality). Methods to calculate the confidence intervals and test the null hypotheses (BACO index = 1) were described and applied in systematically selected clinical trials. Two other preselected trials were included as “positive controls” for being examples of primary composite outcomes disregarded due to inconsistency with the treatment effect on mortality.ResultsThe BACO index values different from one were classified according to whether the use of composite outcomes overestimated (BACO index >1), underestimated (BACO index between zero and ConclusionBACO index can warn that the composite outcome association is stronger, weaker, or even opposite than that of its most critical component.What is new?BACO index is the ratio of the logarithm of the association measure (e.g., RR) of the composite outcome to the logarithm of the association measure of the component endpoint that represents the study target (e.g., mortality).In comparative studies, the BACO index can be used to evaluate the correspondence between the effect on a composite outcome and that on its most critical component.This index could help to preset rules to make decisions for interpretation of clinical studies.A significant BACO should lead to the caution that the association of the composite is stronger (BACO index >1), weaker (BACO index between zero and

It is known that ”standard methods for estimating the causal effect of a time-varying treatment on the mean of a repeated measures outcome (for example, GEE regression) may be biased when there are time-dependent variables that are simultaneously confounders of the effect of interest and are predicted by previous treatment” (Hernan et al. 2002). Inverse-probability of treatment weighted (IPTW) methods are developed in the literature of causal inference. In genetic studies, however, the main interest is to estimate or test the genetic effect rather than the treatment effect. In this work, we describe an IPTW method that provides unbiased estimate for the genetic effect, and discuss how to develop a family-based association test using IPTW for family-based studies. We apply the developed methods to systolic blood pressure data in Framingham Heart Study, where some subjects took antihypertensive treatment during the course of study.

In this paper, we study the statistical issues related to the omission and categorization of important covariates in the context of the Fine–Gray model in randomized-controlled trials with competing risks. We show that the omission of an important covariate from the Fine–Gray model leads to attenuated estimates for treatment effect and loss of proportionality in general. Our simulation studies reveal substantial attenuation in the estimate for treatment effect and the loss of statistical power, while dichotomizing a continuous covariate leads to similar but less pronounced impact. Our results are illustrated using data from a randomized clinical trial of HIV-infected individuals. The relative merits of conducting an adjusted versus an unadjusted analysis of treatment effect in light of both statistical and practical considerations are discussed.

Group sequential designs play an important role in monitoring clinical trials. In this dissertation, we consider the construction of one-sided group sequential designs where the stopping rule includes boundaries for early stopping to accept for futility and to reject for efficacy. The traditional assumption that all patients have the same likelihood of benefiting from the treatment is sometimes unrealistic and it can underestimate the required sample size. This motivates us to power the design for an alternative where the treatment group observations come from a mixture of normal distributions. For the proposed setting, we use standardized test statistics based on sample means. Stopping boundaries and arm size for the design are determined by Type I and Type II error spending equations. The unknown variance case is discussed, and we introduce the group sequential t-test for the mixture setting. With the mixture model, we discuss a more general definition of treatment effect. The maximum likelihood estimator and method of moment estimator for the treatment effect are discussed.

For equivalence trials with survival outcomes, a popular testing approach is the elegant test for equivalence of two survival functions suggested by Wellek (Biometrics 49: 877-881, 1993). This test evaluates whether or not the difference between the true survival curves is practically irrelevant by specifying an equivalence margin on the hazard ratio under the proportional hazards assumption. However, this approach is based on extrapolating the behavior of the survival curves to the whole time axis, whereas in practice survival times are only observed until the end of follow-up. We propose a modification of Welleks test that only addresses equivalence until end of follow-up and derive the large sample properties of this test. Another issue is the proportional hazards assumption which may not be realistic. If this assumption is violated, one may severely misjudge the actual treatment effect with a hazard ratio quantification and wrongly declare equivalence. We suggest a non-parametric test for assessing survival equivalence within the follow-up period. We derive the large sample properties of this test and provide an approximation to the limiting distribution under some mild assumptions on the functional form of the difference between the two survival curves. Both suggestions are investigated by simulation and applied to a clinical trial on survival of gastric cancer patients.

In clinical and epidemiologic studies of time to event, the treatment effect is often of direct interest, and the treatment effect is not constant over time. In this paper, the authors propose an estimator for the cumulative hazard difference under a stratified additive hazards model. The asymptotic properties of the resulting estimator are established, and the finite-sample properties are examined through simulation studies. An application to a liver cirrhosis data set from the Copenhagen Study Group for Liver Diseases is provided.

We consider the problem of estimating the best subgroup and testing for treatment effect in a clinical trial. We define the best subgroup as the subgroup that maximizes a utility function that reflects the trade-off between the subgroup size and the treatment effect. For moderate effect sizes and sample sizes, simpler methods for subgroup estimation worked better than more complex tree- based regression approaches. We propose a three-stage design with a weighted inverse normal combination test to test the hypothesis of no treatment effect across the three stages.

Treatment effect heterogeneity is commonly investigated in meta-analyses to identify if treatment effects vary across studies. When conducting an aggregate level data meta-analysis it is common to describe the magnitude of any treatment effect heterogeneity using the I-squared statistic, which is an intuitive and easily understood concept. The effect of a treatment might also vary across clusters in a cluster randomized trial, or across centres in multi-centre randomized trial, and it can be of interest to explore this at the analysis stage. In cross-over trials and other randomized designs, in which clusters or centres are exposed to both treatment and control conditions, this treatment effect heterogeneity can be identified. Here we derive and evaluate a comparable I-squared measure to describe the magnitude of heterogeneity in treatment effects across clusters or centres in randomized trials. We further show how this methodology can be used to estimate treatment effect heterogeneity in an individual patient data meta-analysis.

Given the unconfoundedness assumption, we propose new nonparametric estimators for the reduced dimensional conditional average treatment effect (CATE) function. In the first stage, the nuisance functions necessary for identifying CATE are estimated by machine learning methods, allowing the number of covariates to be comparable to or larger than the sample size. The second stage consists of a low-dimensional local linear regression, reducing CATE to a function of the covariate(s) of interest. We consider two variants of the estimator depending on whether the nuisance functions are estimated over the full sample or over a hold-out sample. Building on Belloni at al. (2017) and Chernozhukov et al. (2018), we derive functional limit theory for the estimators and provide an easy-to-implement procedure for uniform inference based on the multiplier bootstrap. The empirical application revisits the effect of maternal smoking on a baby's birth weight as a function of the mother's age., 73 pages, fixed the missing reference in the previous version

Cluster-randomized trials (CRTs) of infectious disease preventions often yield correlated, interval-censored data: dependencies may exist between observations from the same cluster, and event occurrence may be assessed only at intermittent study visits. This data structure must be accounted for when conducting interim monitoring and futility assessment for CRTs. In this article, we propose a flexible framework for conditional power estimation when outcomes are correlated and interval-censored. Under the assumption that the survival times follow a shared frailty model, we first characterize the correspondence between the marginal and cluster-conditional survival functions, and then use this relationship to semiparametrically estimate the cluster-specific survival distributions from the available interim data. We incorporate assumptions about changes to the event process over the remainder of the trial-as well as estimates of the dependency among observations in the same cluster-to extend these survival curves through the end of the study. Based on these projected survival functions, we generate correlated interval-censored observations, and then calculate the conditional power as the proportion of times (across multiple full-data generation steps) that the null hypothesis of no treatment effect is rejected. We evaluate the performance of the proposed method through extensive simulation studies, and illustrate its use on a large cluster-randomized HIV prevention trial.

During clinical development of new medicinal products, phase II is one of the most important stages. The first phase II trial is typically a Proof of Concept (PoC) study with limited sample size. To reduce the bias, one needs to discount the phase II estimate of the treatment effect. Under some mild conditions, the estimated assurances are increasing functions of the per group number of patients and the scaled observed treatment effect of the phase II trial for the three cases (no, additive, and multiplicative bias adjustment); and for multiplicative bias adjustment, the estimated assurance is an increasing function of the retention factor. The theoretical assurances are increasing functions of the per group number of patients of the phase II trial and the scaled true treatment effect of the phase III trial. The numerical simulations illustrate the above theoretical results. After obtaining the results of the phase II trial, it is still difficult for the project team (and the company) to make the Go/No Go decision. Finally, in addition to the investment and the statistical points of view, the Go/No Go decision is also affected by many other factors.

Basket trials are a recent and innovative approach in oncological clinical trial design. A basket trial is a type of clinical trial for which eligibility is based on the presence of a specific genomic alteration, irrespective of cancer type. Additionally, basket trials are often used to evaluate the response rate of an investigational therapy across several types of cancer. Recently developed statistical methods for evaluating the response rate in basket trials can be generally categorized into two groups: (a) those that account for the degrees of homogeneity/heterogeneity of response rates among subpopulations, and (b) those using borrowed response rate information across subpopulations to improve the statistical efficiency using Bayesian hierarchical models. In this study, we developed a new basket trial design that accounts for the uncertainties of homogeneity and heterogeneity of response rates among subpopulations using the Bayesian model averaging approach. We demonstrated the utility of the proposed method by comparing our approach against other methods for the two methodological groups using simulated and actual data. On an average, the proposed methods offered an intermediate performance between the BHM-weak and BHM-strong methods. The proposed method would be useful for "signal-finding" basket trials without prior information on the treatment effect of an investigational drug, in part because the proposed method does not require specifications regarding prior distributions of homogeneity response rates among subpopulations.

Recent work has shown that outcomes in clinical trials can be affected by which treatment the trial participants would select if they were allowed to do so, and if they do or do not actually receive that treatment. These influences are known as selection and preference effects, respectively. Unfortunately, they cannot be evaluated in conventional, parallel group trials because patient preferences remain unknown. However, several alternative designs have been proposed, to measure and take account of patient preferences. In this paper, we discuss three preference-based designs (the two-stage, fully randomised, and partially randomised designs). In conventional trials, only the treatment effect is estimable, while the preference-based designs have the potential to estimate some or all of the selection and preference effects. The relative efficiency of these designs is affected by several factors, including the proportion of participants who are undecided about treatments, or who are unable or unwilling to state a preference; the relative preference rate between the treatments being compared, among patients who do have a preference; and the ratio of patients randomised to each treatment. We also discuss the advantages and disadvantages of these designs under different scenarios.

The magnitude of the variation coefficient (CV) is insufficient to validate the quality of the experiment, regardless of the number of treatments, repetitions and effect of treatments. The objective was to develop a new approach to the study of coefficient of variation, as well as evaluations of these nuances with applicability in new scientific research. The study was conducted via computer simulation. The replicates (r) ranged from 2, 3, 4, 5, 10 to 20. The treatment number (t) ranged from t 5, 10, 15, 20, 25 and 30. In each of these combined scenarios we have the variation of 25 different CVs, ranging from 1, 3, 5, 7, ..., 49 to 51 %. It was imposed the variation of 11-1 treatment effects 0, 240, 480, 720, ..., 2000, 2400 kg ha-1, totaling 9,900.00 scenarios. The type I error is statistically invariant in the scenarios studied. With high treatment effect the CV has no implications on the power of the test (1-β). The results obtained in this research reveal that experiments with a high percentage of CV are sufficient to obtain high probabilities of the power of the F test, which do not compromise the complementary analyzes.

The statistical analysis in presence of missing data in any study is challenging. It gets more attention since last few years for clinical trials. There are several reasons for the occurrence of missing data in the crossover trial. However, attempts toward crossover trial data are negligible. This manuscript is dedicated towards development of missing data handling technique for three arms three periods crossover trial.Data obtained from a crossover trial having microarray gene expression values are considered. The gene expression values are considered as outcomes with therapeutic effects. The statistical methodology are explained through Multiple Imputation and Bayesian approach separately. Further, their performance with same data is documented. In Bayesian context, it becomes feasible to perform the causal effect relation jointly with imputation. However, we failed to perform it through mixed effect model jointly. We performed separately Multiple Imputation procedures to overcome the missing values in the dataset and thereafter performed with the mixed effect model to explore the causal effect relation between therapeutic arm on gene expression values.

Single-case experimental designs (SCED) evaluate treatment effects for each participant, but it is difficult to aggregate and quantify treatment effects across SCED participants receiving the same type of treatment. We applied Bayesian analytic procedures to SCED data aggregated across participants that have previously only been applied to large-N and group design studies of treatment effect sizes. For the current study, we defined transient elevated treatment data points as (1) above the range of the last five baseline sessions during the first three sessions of treatment (i.e., extinction burst); (2) within or above the range of baseline after the first three treatment sessions (i.e., recurrence burst); or (3) thinning phase data points above the last three prethinning treatment data points (i.e., thinning burst). Results indicated that the treatment effect sizes remained large regardless of controlling for transient elevated treatment data points. Finally, we examined the effects of reinforcer schedule thinning on estimates of treatment effect size. Results indicated a moderate negative impact of schedule thinning on treatment effect size with a 16% decrease in effect size. Recommendations for research and practice are provided, and the utility of using Bayesian analysis in single-case experimental designs is discussed.

In applied studies, the influence of balance measures on the performance of matching estimators is often taken for granted. This paper considers the performance of different balance measures that have been used in the literature when balance is being optimized. We also propose the use of the entropy measure in assessing balance. To examine the effect of balance measures, we conduct a series of simulation studies where we optimize balance using genetic algorithm (GenMatch). We found that balance measures do influence matching estimates under the GenMatch algorithm. The bias and root-mean-square error (RMSE) of the estimated treatment effect vary with the choice of balance measure. While the performance of different balance measures vary across simulation designs, some pattern did emerge. In an artificial data generating process (DGP) with one covariate, the proposed entropy balance measure has the lowest RMSE. However, in more realistic DGPs with many covariates, the standardized difference in means appear to be a very robust measure of balance, this measure either dominate other measures or come in as a close second in terms of bias or/and RMSE. The implication of these results is that sensitivity of matching estimates to the choice of balance measure should be given greater attention in empirical studies.

Background Exploration and modelling of heterogeneous treatment effects as a function of baseline covariates is an important aspect of precision medicine in randomised controlled trials (RCTs). Randomisation generally guarantees the internal validity of an RCT, but heterogeneity in treatment effect can reduce external validity. Estimation of heterogeneous treatment effects is usually done via a predictive model for individual outcomes, where one searches for interactions between treatment allocation and important patient baseline covariates. However, such models are prone to overfitting and multiple testing and typically demand a transformation of the outcome measurement, for example, from the absolute risk in the original RCT to log-odds of risk in the predictive model. Methods We show how reference classes derived from baseline covariates can be used to explore heterogeneous treatment effects via a two-stage approach. We first estimate a risk score which captures on a single dimension some of the heterogeneity in outcomes of the trial population. Heterogeneity in the treatment effect can then be explored via reweighting schemes along this axis of variation. This two-stage approach bypasses the search for interactions with multiple covariates, thus protecting against multiple testing. It also allows for exploration of heterogeneous treatment effects on the original outcome scale of the RCT. This approach would typically be applied to multivariable models of baseline risk to assess the stability of average treatment effects with respect to the distribution of risk in the population studied. Case study We illustrate this approach using the single largest randomised treatment trial in severe falciparum malaria and demonstrate how the estimated treatment effect in terms of absolute mortality risk reduction increases considerably in higher risk strata. Conclusions ‘Local’ and ‘tilting’ reweighting schemes based on ranking patients by baseline risk can be used as a general approach for exploring, graphing and reporting heterogeneity of treatment effect in RCTs. Trial registration ISRCTN clinical trials registry: ISRCTN50258054. Prospectively registered on 22 July 2005.

In randomized controlled trials of seriously ill patients, death is common and often defined as the primary endpoint. Increasingly, non-mortality outcomes such as functional outcomes are co-primary or secondary endpoints. Functional outcomes are not defined for patients who die, referred to as "truncation due to death", and among survivors, functional outcomes are often unobserved due to missed clinic visits or loss to follow-up. It is well known that if the functional outcomes "truncated due to death" or missing are handled inappropriately, treatment effect estimation can be biased. In this paper, we describe the package idem that implements a procedure for comparing treatments that is based on a composite endpoint of mortality and the functional outcome among survivors. Among survivors, the procedure incorporates a missing data imputation procedure with a sensitivity analysis strategy. A web-based graphical user interface is provided in the idem package to facilitate users conducting the proposed analysis in an interactive and user-friendly manner. We demonstrate idem using data from a recent trial of sedation interruption among mechanically ventilated patients.

This article proposes different tests for treatment effect heterogeneity when the outcome of interest, typically a duration variable, may be right-censored. The proposed tests study whether a policy 1) has zero distributional (average) effect for all subpopulations defined by covariate values, and 2) has homogeneous average effect across different subpopulations. The proposed tests are based on two-step Kaplan-Meier integrals and do not rely on parametric distributional assumptions, shape restrictions, or on restricting the potential treatment effect heterogeneity across different subpopulations. Our framework is suitable not only to exogenous treatment allocation but can also account for treatment noncompliance - an important feature in many applications. The proposed tests are consistent against fixed alternatives, and can detect nonparametric alternatives converging to the null at the parametric $n^{-1/2}$-rate, $n$ being the sample size. Critical values are computed with the assistance of a multiplier bootstrap. The finite sample properties of the proposed tests are examined by means of a Monte Carlo study and an application about the effect of labor market programs on unemployment duration. Open-source software is available for implementing all proposed tests.

Weighted log-rank statistics and recently weighted win loss statistics are often used to test the null hypothesis that the treatment group and the control group have no difference. However, they usually do not provide meaningful treatment effect estimates. This paper studies a few weighted log-rank statistics and weighted win loss statistics that will provide meaningful treatment effect estimates.

In clinical research, longitudinal trials are frequently conducted to evaluate the treatment effect by comparing trends in repeated measurements among different intervention groups. For such longit...

This article assesses the likely generalizability of educational treatment-effect estimates from regression discontinuity designs (RDDs) when treatment assignment is based on academic pretest score...