Background and Objectives: Missing participant data are likely to bias the results of randomized control trials (RCTs) when the reason for missingness is associated with status on the outcome of interest. Unlike dichotomous MPD in RCTs, which have been thoroughly investigated, knowledge regarding continuous MPD in RCTs is much more limited. Our objectives were 1) using an adapted checklist, to assess the reporting quality of simulation studies comparing methods to deal with continuous MPD; 2) identify optimal methods proposed by biostatisticians and tested in simulations studies for continuous MPD in RCTs; 3) evaluate how authors report MPD, and how they plan and conduct analyses to deal with MPD in RCTs. Methods: We conducted two systematic surveys. The first identified methods papers published till 2015 January that compared statistical approaches to deal with continuous MPD in RCTs using at least one simulation. In this sample, we considered both the quality of reporting and the results. The second survey identified a representative sample of individual RCTs published in 2014 in core journals reporting the results of at least one continuous variable addressing a patient-important outcome. Results and conclusion: Our survey identified important limitations in reporting quality of simulation studies that compared statistical approaches to deal with continuous MPD, particularly in the reporting of simulation procedures. Only one of 60 studies reported the random number generator used and none reported starting seeds or failures during simulation. Less then half reported software used to perform simulation (41.7%) or analysis (48.3%), and only 4 (5%) reported justification of number of simulations. When facing continuous MPD in RCTs, results of simulation studies demonstrate that trialists seeking optimal approaches may choose robust regression or mixed models and avoid using last observation caring forward. Continuous MPD frequently occurs in RCTs and the extent is typically substantial (median greater than 10%). Methods sections in trial reports typically do not provide adequate detail on how they dealt with MPD in their primary analysis. Among methods actually implemented to deal with MPD, most authors use only available data, thus excluding MPD from the analysis. Seldom do investigators apply statistical approaches to impute or taking into account of MPD nor conduct sensitivity analysis to address the impact of it. A comprehensive knowledge synthesis summarizing current available statistical approaches and its relative merits, as well as the current used methods in RCTs provide clear implications on how the practise of using methods to handle continuous MPD should shift in individual RCTs. Trialists should use mixed models and robust regressions and avoid using last observation caring forward method. / Thesis / Doctor of Philosophy (PhD)

After reviewing Multifactor Dimensionality Reduction(MDR) and its extensions, an approach to obtain P(larger than 1) risk scores is proposed to predict the continuous outcome for each subject. We study the mean square error(MSE) of dimensionality reduced models fitted with sets of 2 risk scores and investigate the MSE for several special cases of the covariance matrix. A methodology is proposed to select a best set of P risk scores when P is specified a priori. Simulation studies based on true models of different dimensions(larger than 3) demonstrate that the selected set of P(larger than 1) risk scores outperforms the single aggregated risk score generated in AQMDR and illustrate that our methodology can determine a best set of P risk scores effectively. With different assumptions on the dimension of the true model, we considered the preferable set of risk scores from the best set of two risk scores and the best set of three risk scores. Further, we present a methodology to access a set of P risk scores when P is not given a priori. The expressions of asymptotic estimated mean square error of prediction(MSPE) are derived for a 1-dimensional model and 2-dimensional model. In the last main chapter, we apply the methodology of selecting a best set of risk scores where P has been specified a priori to Alzheimer’s Disease data and achieve a set of 2 risk scores and a set of three risk scores for each subject to predict measurements on biomarkers that are crucially involved in Alzheimer’s Disease.