Statistical Methods in Medical Research最新文献

Missing data is a widespread issue in clinical trials, but is particularly problematic for digital health interventions where disengagement is common and outcomes are likely to be missing not at random (MNAR). Trials that use response-adaptive designs need to handle missingness online and not simply at the end of the trial. We propose a novel online imputation strategy which allows previous imputations to be re-imputed given updated estimates of success probabilities. We additionally consider: (i) truncation of deterministic algorithms to prevent extreme realised treatment imbalance and (ii) changing the random component of semi-randomised algorithms. Through a simulation study based on a trial for a digital smoking cessation intervention, we illustrate how the strategy for handling missing responses can affect the exploration-exploitation tradeoff and the bias of the estimated success probabilities at the end of the trial. In the settings explored, we found that the exploration-exploitation tradeoff is affected particularly when arms have very different rates of missingness and we identified combinations of response-adaptive designs and missingness strategies that are particularly problematic. Further, we show that estimated success probabilities at the end of the trial can be biased not only due to optimistic sampling, but potentially also due to an MNAR missingness mechanism.

Multivariate imputation using chained equations is a popular algorithm for imputing missing data that entails specifying multivariable models through conditional distributions. Two standard imputation methods for imputing missing continuous variables are parametric imputation using a linear model and predictive mean matching. The default methods for imputing missing categorical variables are parametric imputation using multinomial logistic regression and ordinal logistic regression for imputing nominal and ordinal categorical variables, respectively. There is a paucity of research into the relative computational burden and the quality of statistical inferences when using predictive mean matching versus parametric imputation for imputing missing non-binary categorical variables. We used simulations to compare the performance of predictive mean matching with that of multinomial logistic regression and ordinal logistic regression for imputing categorical variables when the analysis model of scientific interest was a logistic or linear regression model. We varied the sample size (N = 500, 1000, 2500, and 5000), the rate of missing data (5%-50% in increments of 5%), and the number of levels of the categorical variable (3, 4, 5, and 6). In general, the performance of predictive mean matching compared very favorably to that of multinomial or ordinal logistic regression for imputing categorical variables when the analysis model was a logistic or linear regression model. This was true across a range of scenarios defined by sample size and the rate of missing data. Furthermore, the use of predictive mean matching was substantially faster, by a factor of 2-6. In conclusion, predictive mean matching can be used to impute categorical variables. The use of predictive mean matching to impute missing non-binary categorical variables substantially reduces computer processing time when conducting multiple imputation.

Cluster randomized trials are widely used in healthcare research for the evaluation of intervention strategies. Beyond estimating the average treatment effect, it is often of interest to assess whether the treatment effect varies across subgroups. While conventional methods based on tests of interaction terms between treatment and covariates can be used to detect treatment effect heterogeneity in cluster randomized trials, they typically rely on parametric assumptions that may not hold in practice. Adapting existing permutation tests from individually randomized trials, however, requires conceptual clarification and modification due to the multiple possible interpretations of treatment effect heterogeneity in the cluster randomized trial context. In this work, we develop variations of permutation tests and clarify key causal definitions in order to assess treatment effect heterogeneity in cluster randomized trials. Our procedure enables investigators to simultaneously test for effect modification across a large number of covariates, while maintaining nominal type I error rates and reasonable power in simulation studies. In the Pain Program for Active Coping and Training (PPACT) study, the proposed methods are able to detect treatment effect heterogeneity that was not identified by conventional methods assessing treatment-covariate interactions.

Lithium is recommended as a first line treatment for patients with bipolar disorder. However, only certain patients show a good response to the drug, and the variability and tolerability of lithium response are poorly understood. Greater precision in the early identification of individuals who are likely to respond well to lithium is a significant unmet clinical need. We create optimal designs to better understand the pharmacokinetic exposition of lithium for patients with and without a genetic covariate. From a Fisher information matrix based method, we find different optimal designs for estimating various parameters in a complicated pharmacokinetics/pharmacodynamics nonlinear mixed effects model with multiple physician specified constraints. Our approach uses flexible state-of-the-art metaheuristics to find various types of efficient designs, including multiple-objective optimal designs that can balance the competitiveness of the objectives and deliver higher efficiencies for more important objectives. Results from this article will be used as part of a broader study to implement efficient designs to better understand the exposition of sustained-release lithium in patients with bipolar disorder.

In clinical research of cancer therapy for rare mutations, trial designs must be adapted to accommodate the typically small sample sizes, and single-arm and basket trials have gained prominence. In this paper, we apply principles of Bayesian hierarchical methods and multilevel network meta-regression to propose a model for a pairwise population-adjusted unanchored indirect comparison of cancer therapies in different tumor types with borrowing of pan-tumor information. An individual-level regression model is defined for the single-arm trial of the intervention for which we have individual patient data. The aggregate data of the other trial for the competing intervention are fitted by integrating the covariate effects at the individual level over its covariate distribution to form the aggregate likelihood. To improve the estimation of the tumor type-specific relative treatment effects, we assume exchangeability reflecting the belief of a pan-tumor effect. The method is illustrated with a case study of adagrasib versus sotorasib in previously treated KRAS^G12C-mutated advanced/metastatic tumors: non-small cell lung cancer (NSCLC), colorectal cancer (CRC), and pancreatic ductal adenocarcinoma (PDAC). Adagrasib was associated with a greater tumor response than sotorasib according to the analyses: The odds ratios were 1.87 (1.21-2.84) for NSCLC; 2.08 (1.22-3.93) for CRC; and 2.02 (1.14-4.05) for PDAC. The analysis illustrated that a reasonably conservative assumption about the degree of similarity can result in more meaningful and interpretable findings. The proposed model allows for population adjustment and information sharing across tumor types when performing an unanchored indirect comparison of interventions for which it is believed a pan-tumor effect holds.

Dose-finding trials are designed to identify a safe and potentially effective drug dose and schedule during the early phase of clinical trials. Historically, Bayesian adaptive dose-escalation methods in Phase I trials in cancer have mainly focussed on toxicity endpoints rather than efficacy endpoints. This is partly because efficacy readouts are often not available soon enough for dose escalation decisions. In the last decade, 'liquid biopsy' technologies have been developed, which may provide a readout of treatment response much earlier than conventional endpoints. This paper develops a novel design that uses a biomarker, circulating tumour DNA (ctDNA), with toxicity and activity outcomes in dose-finding studies. We compare the proposed approach based on repeated ctDNA measurement with existing Bayesian adaptive approaches under various scenarios of dose-toxicity, dose-efficacy relationship, and trajectories of regular ctDNA values over time. Simulation results show that the proposed approach can yield significantly shorter trial duration and may improve identification of the target dose. In addition, this approach has the potential to minimise the time individual patients spend on potentially inactive trial therapies. Using two different dose-finding designs, we demonstrate that the way we incorporate biomarker information is broadly applicable across different dose-finding designs and yields notable benefit in both cases.

There are sundry practical situations in which paired count variables are correlated, thus requiring a joint estimation method. In this article, we introduce a flexible class of bivariate mixed Poisson regression models, which settle into an exponential-family (EF) distributed component for unobserved heterogeneity. The proposed bivariate mixed Poisson models deal with the phenomenon of overdispersion, typical of count data, and have flexibility in terms of the correlation structure. Thus, this novel class of models has a distinct advantage over the most widely used models because it captures both positive and negative correlations in the count data. Under the bivariate mixed Poisson model, inference of the parameters is conducted through the maximum likelihood method. Monte Carlo studies on assessing the finite-sample performance of the estimators of the parameters are presented. Furthermore, we employ a likelihood ratio statistic for testing the significance of certain sources of correlation and evaluate its performance via simulation studies. Moreover, model adequacy is addressed by using simulated envelopes for residual analysis, and also a randomized probability integral transformation for calibration model control. The proposed bivariate mixed Poisson model is considered for analyzing a healthcare dataset from the Australian Health Survey, where our aim is to study the association between the number of consultations with a doctor and the number of non-prescribed drug intake.

Plasma donation plays a critical role in modern medicine by providing lifesaving treatments for patients with a wide range of conditions like bleeding disorders, immune deficiencies, and infections. Evaluation of devices used to collect blood plasma from donors is essential to ensure donor safety. We consider the design of plasma donation trials when the goal is to assess the safety of a new device on the response to transfusions compared to the standard device. A unique feature is that the number of donations per donor varies substantially so some individuals contribute more information and others less. The sample size formula is derived to ensure power requirements are met when analyses are based on generalized estimating equations and robust variance estimation. Strategies for interim monitoring based on group sequential designs using alpha spending functions are developed based on a robust covariance matrix for estimates of treatment effect over successive analyses. The design of a plasma donation study is illustrated where the focus is on assessing the safety of a new device with serious hypotensive adverse events as the primary outcome.

In receiver operating characteristicROC analysis, the area under the ROC curve (AUC) is a popular one number summary of the discriminatory accuracy of a diagnostic test. AUC measures the overall diagnostic accuracy of a test but fails to account for the effect of covariates when covariates are present and associated with the test results. Adjustment for covariate effects can greatly improve the diagnostic accuracy of a test. In this paper, using information provided by the influence function, empirical likelihood (EL) methods are proposed for inferences of AUC in presence of covariates. For parameters in the AUC regression model, it is shown that the asymptotic distribution of the influence function-based empirical log-likelihood ratio statistic is a standard chi-square distribution. Hence, confidence regions for the regression parameters can be obtained without any variance estimation. Simulation studies are conducted to compare the finite sample performances of the proposed EL based methods with the existing normal approximation (NA) based method in the AUC regression. Simulation results indicate that the bootstrap-calibrated influence function-based empirical likelihood (BIFEL ) confidence region outperforms the NA-based confidence region in terms of coverage probability. We also propose an interval estimation method for the covariate-adjusted AUC based on the BIFEL confidence region. Finally, we illustrate the recommended method with a real prostate-specific antigen data example.

Saved time is used in Alzheimer's disease (AD) trials as an easy interpretation of the treatment benefit to communicate with patients, family members, and caregivers. The projection approach is frequently applied to estimate saved time and its confidence interval (CI) by using the placebo or treatment disease progression curves. The estimated standard error of saved time by using these existing methods does not account for the correlation between outcomes. In addition, there was no closed-form CI for researchers to use in practice. To fill this critical gap, we derive the closed-form CI for saved time estimated from the placebo or treatment disease progression curves. We compare them with regard to coverage probability and interval width under various disease progression patterns that are commonly observed in AD symptomatic therapy and disease-modifying therapy trials. Data from the phase 3 donanemab trials are used to illustrate the application of the new CI methods.