Statistics in medicine
-
Statistics in medicine · Jan 2014
Bias associated with using the estimated propensity score as a regression covariate.
The use of propensity score methods to adjust for selection bias in observational studies has become increasingly popular in public health and medical research. A substantial portion of studies using propensity score adjustment treat the propensity score as a conventional regression predictor. Through a Monte Carlo simulation study, Austin and colleagues. investigated the bias associated with treatment effect estimation when the propensity score is used as a covariate in nonlinear regression models, such as logistic regression and Cox proportional hazards models. ⋯ Instead of specifying a known parametric propensity score model, we generate the data by considering various degrees of overlap of the covariate distributions between treated and control groups. Propensity score matching excels when the treated group is contained within a larger control pool, while the model-based adjustment may have an edge when treated and control groups do not have too much overlap. Overall, adjusting for the propensity score through stratification or matching followed by regression or using splines, appears to be a good practical strategy.
-
Statistics in medicine · Dec 2013
Visualizing the flow of evidence in network meta-analysis and characterizing mixed treatment comparisons.
Network meta-analysis techniques allow for pooling evidence from different studies with only partially overlapping designs for getting a broader basis for decision support. The results are network-based effect estimates that take indirect evidence into account for all pairs of treatments. The results critically depend on homogeneity and consistency assumptions, which are sometimes difficult to investigate. ⋯ The graphical display and the measures are illustrated for two published network meta-analyses. In these applications, the proposed methods are seen to render transparent the process of data pooling in mixed treatment comparisons. They can be expected to be more generally useful for guiding and facilitating the validity assessment in network meta-analysis.
-
Statistics in medicine · Nov 2013
Designs for randomized phase II clinical trials with two treatment arms.
The most common primary statistical end point of a phase II clinical trial is the categorization of a patient as either a 'responder' or 'nonresponder'. The primary objective of typical randomized phase II anticancer clinical trials is to evaluate experimental treatments that potentially will increase response rate over a historical baseline and select one to consider for further study. ⋯ We develop a program to compute these error rates and powers exactly, and we provide many design examples to satisfy pre-fixed requirements on error rates and powers. Finally, we apply our method to a randomized phase II trial in patients with relapsed non-Hodgkin's disease.
-
Statistics in medicine · Oct 2013
Estimation of causal mediation effects for a dichotomous outcome in multiple-mediator models using the mediation formula.
Mediators are intermediate variables in the causal pathway between an exposure and an outcome. Mediation analysis investigates the extent to which exposure effects occur through these variables, thus revealing causal mechanisms. In this paper, we consider the estimation of the mediation effect when the outcome is binary and multiple mediators of different types exist. ⋯ The results also show that the power to detect a nonzero total mediation effect increases as the correlation coefficient between two mediators increases, whereas power for individual mediation effects reaches a maximum when the mediators are uncorrelated. We illustrate our approach by applying it to a retrospective cohort study of dental caries in adolescents with low and high socioeconomic status. Sensitivity analysis is performed to assess the robustness of conclusions regarding mediation effects when the assumption of no unmeasured mediator-outcome confounders is violated.
-
Statistics in medicine · Sep 2013
A simple, flexible, and effective covariate-adaptive treatment allocation procedure.
We present a method for allocating treatment when subjects arrive in sequence. Based on the theory of propensity scores more commonly used in observational studies, the method balances both discrete and continuous covariates without assuming a model for the outcome. Although we allow for a number of possible specifications, we explore some specific instances in depth. ⋯ We also investigate the properties of selected randomized versions with respect to both optimality and selection bias. We conclude with an application to a pilot study on weight loss. The proposed method is shown to be robust to the number of covariates balanced and the marginal and joint distributions of those covariates.