Statistical Methods in Medical Research最新文献

The cause-specific hazard Cox model is widely used in analyzing competing risks survival data, and the partial likelihood method is a standard approach when survival times contain only right censoring. In practice, however, interval-censored survival times often arise, and this means the partial likelihood method is not directly applicable. Two common remedies in practice are (i) to replace each censoring interval with a single value, such as the middle point; or (ii) to redefine the event of interest, such as the time to diagnosis instead of the time to recurrence of a disease. However, the mid-point approach can cause biased parameter estimates. In this article, we develop a penalized likelihood approach to fit semi-parametric cause-specific hazard Cox models, and this method is general enough to allow left, right, and interval censoring times. Penalty functions are used to regularize the baseline hazard estimates and also to make these estimates less affected by the number and location of knots used for the estimates. We will provide asymptotic properties for the estimated parameters. A simulation study is designed to compare our method with the mid-point partial likelihood approach. We apply our method to the Aspirin in Reducing Events in the Elderly (ASPREE) study, illustrating an application of our proposed method.

Linear mixed models are commonly used in analyzing stepped-wedge cluster randomized trials. A key consideration for analyzing a stepped-wedge cluster randomized trial is accounting for the potentially complex correlation structure, which can be achieved by specifying random-effects. The simplest random effects structure is random intercept but more complex structures such as random cluster-by-period, discrete-time decay, and more recently, the random intervention structure, have been proposed. Specifying appropriate random effects in practice can be challenging: assuming more complex correlation structures may be reasonable but they are vulnerable to computational challenges. To circumvent these challenges, robust variance estimators may be applied to linear mixed models to provide consistent estimators of standard errors of fixed effect parameters in the presence of random-effects misspecification. However, there has been no empirical investigation of robust variance estimators for stepped-wedge cluster randomized trials. In this article, we review six robust variance estimators (both standard and small-sample bias-corrected robust variance estimators) that are available for linear mixed models in R, and then describe a comprehensive simulation study to examine the performance of these robust variance estimators for stepped-wedge cluster randomized trials with a continuous outcome under different data generators. For each data generator, we investigate whether the use of a robust variance estimator with either the random intercept model or the random cluster-by-period model is sufficient to provide valid statistical inference for fixed effect parameters, when these working models are subject to random-effect misspecification. Our results indicate that the random intercept and random cluster-by-period models with robust variance estimators performed adequately. The CR3 robust variance estimator (approximate jackknife) estimator, coupled with the number of clusters minus two degrees of freedom correction, consistently gave the best coverage results, but could be slightly conservative when the number of clusters was below 16. We summarize the implications of our results for the linear mixed model analysis of stepped-wedge cluster randomized trials and offer some practical recommendations on the choice of the analytic model.

Individualized treatment rules inform tailored treatment decisions based on the patient's information, where the goal is to optimize clinical benefit for the population. When the clinical outcome of interest is survival time, most of current approaches typically aim to maximize the expected time of survival. We propose a new criterion for constructing Individualized treatment rules that optimize the clinical benefit with survival outcomes, termed as the adjusted probability of a longer survival. This objective captures the likelihood of living longer with being on treatment, compared to the alternative, which provides an alternative and often straightforward interpretation to communicate with clinicians and patients. We view it as an alternative to the survival analysis standard of the hazard ratio and the increasingly used restricted mean survival time. We develop a new method to construct the optimal Individualized treatment rule by maximizing a nonparametric estimator of the adjusted probability of a longer survival for a decision rule. Simulation studies demonstrate the reliability of the proposed method across a range of different scenarios. We further perform data analysis using data collected from a randomized Phase III clinical trial (SWOG S0819).

A useful parametric specification for the expected value of an epidemiological process is revived, and its statistical and empirical efficacy are explored. The Richards' curve is flexible enough to adapt to several growth phenomena, including recent epidemics and outbreaks. Here, two different estimation methods are described. The first, based on likelihood maximisation, is particularly useful when the outbreak is still ongoing and the main goal is to obtain sufficiently accurate estimates in negligible computational run-time. The second is fully Bayesian and allows for more ambitious modelling attempts such as the inclusion of spatial and temporal dependence, but it requires more data and computational resources. Regardless of the estimation approach, the Richards' specification properly characterises the main features of any growth process (e.g. growth rate, peak phase etc.), leading to a reasonable fit and providing good short- to medium-term predictions. To demonstrate such flexibility, we show different applications using publicly available data on recent epidemics where the data collection processes and transmission patterns are extremely heterogeneous, as well as benchmark datasets widely used in the literature as illustrative.

For personalized medicine, we propose a general method of evaluating the potential performance of an individualized treatment rule in future clinical applications with new patients. We focus on rules that choose the most beneficial treatment for the patient out of two active (nonplacebo) treatments, which the clinician will prescribe regularly to the patient after the decision. We develop a measure of the individualization potential (IP) of a rule. The IP compares the expected effectiveness of the rule in a future clinical individualization setting versus the effectiveness of not trying individualization. We illustrate our evaluation method by explaining how to measure the IP of a useful type of individualized rules calculated through a new parametric interaction model of data from parallel-group clinical trials with continuous responses. Our interaction model implies a structural equation model we use to estimate the rule and its IP. We examine the IP both theoretically and with simulations when the estimated individualized rule is put into practice in new patients. Our individualization approach was superior to outcome-weighted machine learning according to simulations. We also show connections with crossover and N-of-1 trials. As a real data application, we estimate a rule for the individualization of treatments for diabetic macular edema and evaluate its IP.

Cluster randomized crossover and stepped wedge cluster randomized trials are two types of longitudinal cluster randomized trials that leverage both the within- and between-cluster comparisons to estimate the treatment effect and are increasingly used in healthcare delivery and implementation science research. While the variance expressions of estimated treatment effect have been previously developed from the method of generalized estimating equations for analyzing cluster randomized crossover trials and stepped wedge cluster randomized trials, little guidance has been provided for optimal designs to ensure maximum efficiency. Here, an optimal design refers to the combination of optimal cluster-period size and optimal number of clusters that provide the smallest variance of the treatment effect estimator or maximum efficiency under a fixed total budget. In this work, we develop optimal designs for multiple-period cluster randomized crossover trials and stepped wedge cluster randomized trials with continuous outcomes, including both closed-cohort and repeated cross-sectional sampling schemes. Local optimal design algorithms are proposed when the correlation parameters in the working correlation structure are known. MaxiMin optimal design algorithms are proposed when the exact values are unavailable, but investigators may specify a range of correlation values. The closed-form formulae of local optimal design and MaxiMin optimal design are derived for multiple-period cluster randomized crossover trials, where the cluster-period size and number of clusters are decimal. The decimal estimates from closed-form formulae can then be used to investigate the performances of integer estimates from local optimal design and MaxiMin optimal design algorithms. One unique contribution from this work, compared to the previous optimal design research, is that we adopt constrained optimization techniques to obtain integer estimates under the MaxiMin optimal design. To assist practical implementation, we also develop four SAS macros to find local optimal designs and MaxiMin optimal designs.

Prognostic biomarkers for survival outcomes are widely used in clinical research and practice. Such biomarkers are often evaluated using a C-index as well as quantities based on time-dependent receiver operating characteristic curves. Existing methods for their evaluation generally assume that censoring is uninformative in the sense that the censoring time is independent of the failure time with or without conditioning on the biomarker under evaluation. With focus on the C-index and the area under a particular receiver operating characteristic curve, we describe and compare three estimation methods that account for informative censoring based on observed baseline covariates. Two of them are straightforward extensions of existing plug-in and inverse probability weighting methods for uninformative censoring. By appealing to semiparametric theory, we also develop a doubly robust, locally efficient method that is more robust than the plug-in and inverse probability weighting methods and typically more efficient than the inverse probability weighting method. The methods are evaluated and compared in a simulation study, and applied to real data from studies of breast cancer and heart failure.

Bounded count response data arise naturally in health applications. In general, the well-known beta-binomial regression model form the basis for analyzing this data, specially when we have overdispersed data. Little attention, however, has been given to the literature on the possibility of having extreme observations and overdispersed data. We propose in this work an extension of the beta-binomial regression model, named the beta-2-binomial regression model, which provides a rather flexible approach for fitting a regression model with a wide spectrum of bounded count response data sets under the presence of overdispersion, outliers, or excess of extreme observations. This distribution possesses more skewness and kurtosis than the beta-binomial model but preserves the same mean and variance form of the beta-binomial model. Additional properties of the beta-2-binomial distribution are derived including its behavior on the limits of its parametric space. A penalized maximum likelihood approach is considered to estimate parameters of this model and a residual analysis is included to assess departures from model assumptions as well as to detect outlier observations. Simulation studies, considering the robustness to outliers, are presented confirming that the beta-2-binomial regression model is a better robust alternative, in comparison with the binomial and beta-binomial regression models. We also found that the beta-2-binomial regression model outperformed the binomial and beta-binomial regression models in our applications of predicting liver cancer development in mice and the number of inappropriate days a patient spent in a hospital.

The mixture of probabilistic regression models is one of the most common techniques to incorporate the information of covariates into learning of the population heterogeneity. Despite its flexibility, unreliable estimates can occur due to multicollinearity among covariates. In this paper, we develop Liu-type shrinkage methods through an unsupervised learning approach to estimate the model coefficients in the presence of multicollinearity. We evaluate the performance of our proposed methods via classification and stochastic versions of the expectation-maximization algorithm. We show using numerical simulations that the proposed methods outperform their Ridge and maximum likelihood counterparts. Finally, we apply our methods to analyze the bone mineral data of women aged 50 and older.