Statistics in Medicine最新文献

Independent binomially distributed data arise in many contexts, such as clinical trials, quality control monitoring, and stratified sampling. Moreover, the scope is much larger because multinomially distributed data from a 2 by $k$ contingency table can be viewed as $k$ conditionally independent binomial random variables. A standard approach is to use Fisher's conditional test to test equality of the $k$ underlying success probabilities. However, researchers often want to know where the important pairwise differences are. Thus, the closed method of pairwise comparisons is here combined with unconditional exact tests for 2 by 2 tables and Fisher's conditional test for larger tables to get $p$ values exhibiting strong control of the Family-Wise Error Rate and excellent power properties. In clinical trials, studies are often multicenter, but the results here pertain only to single-site studies.

Missing covariates are a common challenge when applying an existing logistic regression model to new or external datasets, particularly in the context of model updating. While regression calibration and model updating methods have been developed to address such partial data availability, each has limitations in terms of bias, variance, and sensitivity to model misspecification. In this study, we propose a surrogate-calibrated updating (SCU) method that integrates calibration and updating approaches to improve the efficiency and reliability of coefficient estimation in the presence of missing covariates. The SCU method leverages surrogate covariates-variables that are routinely available across old and new datasets and correlate with the missing covariates-and applies a weighted averaging scheme that combines information from both fully observed and partially observed data sources. This approach mitigates bias while reducing variance, offering a practical and robust alternative to existing methods in population updating setting. We provide a theoretical justification and derive the corresponding estimators and variances. Simulation studies demonstrate the method's favorable performance under various scenarios, including the case with model misspecification. The SCU method is further illustrated using data from the Framingham Heart Study, where diabetes history serves as a surrogate for partially observed glucose levels in assessing cardiovascular disease risk. JEL Classification: C13, C18, C35.

Modeling dynamic heterogeneity is essential for revealing the distinct longitudinal trajectories of individual change. Dynamic heterogeneity analysis of semi-continuous longitudinal data is commonly difficult due to the semi-continuity of longitudinal responses. The hidden semi-Markov model is a powerful tool that can reveal the longitudinal dependency structure and the dynamic heterogeneity of the observation process by introducing the sojourn time distribution. To address the challenge of modeling dynamic heterogeneity in semi-continuous longitudinal data, this study develops a two-part hidden semi-Markov mixed-effects model. The proposed model mainly consists of two parts: a discrete binary indicator model to estimate the probability of a zero outcome for the semi-continuous longitudinal response, and a continuous hidden semi-Markov model to fit the positive values of semi-continuous longitudinal responses. In order to accurately obtain the state of each individual at different observation times, a set of likelihood ratio test state iteration algorithms is developed. Bayesian methods are used to estimate the regression coefficients and state parameters of the proposed model. The proposed methodology is applied to analyze the dataset of the Health and Retirement Study conducted by the University of Michigan. Simulation studies are conducted to assess the flexibility of the proposed model under various scenarios.

Improving patients' quality of life (QoL) is one of the primary goals of palliative care clinical trials. However, a significant challenge in this area is the "truncation by death problem," where QoL data cannot be observed after a patient dies, potentially introducing bias into statistical analyses. Understanding the impact of truncation by death when estimating the association between QoL and exposure or treatment is essential, especially when a relatively large proportion of subjects die during a study. To address this issue, we propose a Bayesian joint modeling framework that considers dependencies at both the individual and cluster levels while examining longitudinal QoL trajectories and survival outcomes simultaneously. This approach builds on existing joint modeling methods by incorporating cluster-level random effects. We model QoL on a retrospective scale relative to the time of death, while linking survival via both the subject and cluster-level random effects. The longitudinal sub-model also allows for flexible, non-linear QoL trajectories, which are modeled using penalized regression splines. For the survival sub-model, we use a proportional hazards frailty model with a Weibull baseline hazard. The model is estimated using a Bayesian framework, implemented via Markov Chain Monte Carlo (MCMC) sampling. To evaluate the performance of our method, we conducted a comprehensive simulation study including scenarios with different numbers of clusters. We also show results from applying this novel methodology to data from the Reducing End of Life Symptoms with Touch (REST) study.

Stepped-wedge cluster randomized trials (SW-CRTs) are one-way crossover trials that randomize clusters (i.e., groups) of individuals to the time point (period) at which an intervention is introduced into the cluster. In these designs, the intervention under evaluation is introduced into all of the clusters by the end of the study in a series of "steps." Analysis of SW-CRTs using marginal models provides a population-averaged interpretation of the estimated intervention effect and flexible specification of the within-cluster, marginal pairwise association structure; the latter has practical application in reporting intraclass (i.e., pairwise) correlations and calculating power for CRTs. Despite these features, use of marginal modeling of SW-CRTs has been mostly limited to applications with working independence and simple exchangeable correlation structures that are suboptimal for multi-period CRTs when correlation among responses decays over time. However, there have been many methodological developments in marginal modeling of SW-CRTs over the past fifteen years, particularly on (i) multi-parameter, within-cluster correlation structures; (ii) paired generalized estimating equations (GEE) for simultaneous estimation of mean and correlation parameters with standard errors; and, when the number of clusters is small, (iii) corrections to reduce the bias of variance estimators, and that of correlation estimates using matrix-adjusted estimating equations (MAEE). The goal of the current tutorial is to survey these newer developments and to provide case studies to enable applied researchers to implement GEE/MAEE for marginal model analysis of SW-CRTs, with application to both cohorts and designs with repeated cross-sectional samples. The methods are also applicable to multi-period, parallel-arm and cluster-crossover CRTs.

One major bias source in causal inference for clinical trials is unmeasured confounding. We propose an innovative, practical Bayesian modeling approach to adjust for unmeasured confounding effects and obtain precise causal average treatment effect estimates for two-arm randomized controlled clinical trials. This approach includes model reparameterization and an iterative algorithm, with a causal inference framework incorporated with unmeasured confounders and related statistical distributions. Model non-identifiability resulting from adjusting for unmeasured confounding is a major inferential problem. Reparameterization transforms one or multiple unmeasured confounders into a single reparameterized unmeasured confounder and can remove model non-identifiability from the model specification of unmeasured confounders. The iterative algorithm consists of detailed steps for inference after model reparameterization and can remove model non-identifiability from prior sensitivity to unmeasured confounders. It includes iterating the prior distribution of the reparameterized unmeasured confounder by certain rules, aggregating posterior means and variances over different prior choices, and obtaining posterior estimates for the average treatment effect. Its essential idea is to make unreliable prior information on unmeasured confounders as close to data information as possible. Compared with usual methods, our approach produces robust effect estimates and correctly concludes statistical significance. From an example using real clinical data, this approach effectively adjusts for confounding effects when we do not adjust for measured confounders. Our approach is also generalizable to other clinical study designs and may be beneficial to applications where data collection is difficult for certain variables or causal relationships are not well understood.

This paper proposes a propensity score (PS)-based stratified win ratio method to address challenges of small patient populations in clinical trials, especially for rare or pediatric diseases, by incorporating external control data. Our approach enhances traditional win ratio analysis by leveraging PS stratification to account for heterogeneity between the current and external studies. Additionally, down-weighting based on the overlapping coefficient of PS distributions of current treatment and external control groups further mitigates the patient bias due to heterogeneity. Simulations show significant improvements in statistical power for detecting treatment effects within the composite endpoint combining continuous and time-to-event components, over nonborrowing and pooling methods, with utilizing Mantel-Haenszel (MH)-type weights achieving the highest power. The proposed methods are also applied to an amyotrophic lateral sclerosis (ALS) study incorporating the external control arm from a prior ALS trial. The proposed PS-based stratified win ratio method thus provides a rigorous framework for borrowing external data and analyzing composite endpoints with limited patient availability.

A dynamic treatment regime (DTR) is a sequence of treatment decision rules tailored to an individual's evolving status over time. In precision medicine, much focus has been placed on finding an optimal DTR which, if followed by everyone in the population, would yield the best outcome on average; and extensive investigations have been conducted from both methodological and applied standpoints. The purpose of this tutorial is to provide readers who are interested in optimal DTRs with a systematic, detailed, but accessible introduction, including the formal definition and formulation of this topic within the framework of causal inference, identification assumptions required to link the causal quantity of interest to the observed data, existing statistical models and estimation methods for learning the optimal regime from the data, and application of these methods to both simulated and real data.

Seasonality plays a crucial role in the transmission dynamics of many infectious diseases, contributing to periodic fluctuations in disease incidence. The previously developed geographically dependent individual-level model (GD-ILM) has been effective in modeling infectious diseases, but does not incorporate seasonal effects, limiting its ability to capture seasonal trends. In this study, we extend the GD-ILM by introducing a seasonally varying transmission component, allowing the model to account for periodic fluctuations in infection risk. Our approach integrates a seasonally forced infection kernel to model periodic changes in transmission rates over time, leading to a novel spatiotemporal kernel. To facilitate efficient and reliable parameter estimation in this high-dimensional setting, we employ the Monte Carlo expectation conditional maximization algorithm. We apply our model to individual-level influenza A data from Manitoba, Canada, examining spatial and seasonal infection patterns to identify high-risk regions and periods, and thus informing targeted intervention strategies. The proposed model's performance is further validated through comprehensive simulation studies. Simulation results confirm that models omitting seasonal components lead to biased spatial parameter estimates under various disease prevalence conditions. To support reproducibility and practical application, we developed the SeasEpi R package publicly available on the comprehensive R archive network (CRAN), which implements the seasonal GD-ILM framework and provides tools for model fitting, simulation, and evaluation. The seasonal GD-ILM offers a more accurate framework for modeling infectious disease transmission by integrating both spatial and seasonal dynamics. It supports more accurate risk assessment and enhances public health responses by enabling timely and location-specific interventions based on seasonal transmission patterns.

In biomedical research, understanding the dynamic relationships between two binary variables over time is crucial. Our study enhances this understanding by employing longitudinal analysis to introduce measures such as the bivariate time-varying odds ratio and relative risk. These metrics adeptly quantify evolving associations and effectively address the complexities involved in estimating variables recorded at disparate times. We have developed a nonparametric approach specifically designed for longitudinal samples that vary in their measurement timelines, demonstrating its applicability to both concurrent and nonconcurrent sampling scenarios. Additionally, in studies where end-of-life considerations are prevalent, missing data can significantly skew results. To mitigate this, we implemented a model that accounts for missingness and developed an inverse-probability weighting method that has been validated through simulation studies to correct biases effectively. By applying our methodology to the Framingham Heart Study, we investigated the temporal changes in the association of hypertension among mothers and daughters over a 45-year span. This application not only underscores the versatility of our approach but also provides valuable insights into long-term health trends within families.