Dataset Information

Decision-making with multiple correlated binary outcomes in clinical trials.

ABSTRACT: Clinical trials often evaluate multiple outcome variables to form a comprehensive picture of the effects of a new treatment. The resulting multidimensional insight contributes to clinically relevant and efficient decision-making about treatment superiority. Common statistical procedures to make these superiority decisions with multiple outcomes have two important shortcomings, however: (1) Outcome variables are often modeled individually, and consequently fail to consider the relation between outcomes; and (2) superiority is often defined as a relevant difference on a single, on any, or on all outcome(s); and lacks a compensatory mechanism that allows large positive effects on one or multiple outcome(s) to outweigh small negative effects on other outcomes. To address these shortcomings, this paper proposes (1) a Bayesian model for the analysis of correlated binary outcomes based on the multivariate Bernoulli distribution; and (2) a flexible decision criterion with a compensatory mechanism that captures the relative importance of the outcomes. A simulation study demonstrates that efficient and unbiased decisions can be made while Type I error rates are properly controlled. The performance of the framework is illustrated for (1) fixed, group sequential, and adaptive designs; and (2) non-informative and informative prior distributions.

SUBMITTER: Kavelaars X

PROVIDER: S-EPMC7682528 | biostudies-literature | 2020 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Decision-making with multiple correlated binary outcomes in clinical trials.

Kavelaars Xynthia X Mulder Joris J Kaptein Maurits M

Statistical methods in medical research 20200716 11

Clinical trials often evaluate multiple outcome variables to form a comprehensive picture of the effects of a new treatment. The resulting multidimensional insight contributes to clinically relevant and efficient decision-making about treatment superiority. Common statistical procedures to make these superiority decisions with multiple outcomes have two important shortcomings, however: (1) Outcome variables are often modeled individually, and consequently fail to consider the relation between ou ...[more]

PMID: 32672498

Similar Datasets

Project description:BackgroundThere is an increasing body of evidence supporting a more flexible approach in clinical data requirements for the approval of more complex biosimilar substances such as monoclonal antibodies (mAbs).ObjectiveThe aim of this paper is to further analyse the role of quality/chemistry, manufacturing and controls (CMC) and clinical data for the conclusion on biosimilarity and the decision on marketing authorisation (MA).MethodsIn the present study, we analysed the MA applications (MAAs) of all 33 mAbs and three fusion proteins evaluated by the European Medicines Agency (EMA) between July 2012 and November 2022 with special emphasis on all submitted rituximab (four products) and trastuzumab (seven products) biosimilar candidates, including withdrawn applications. For the two withdrawn applications, the comparative efficacy trials suggested biosimilarity, but the quality/CMC package was not accepted by EMA. We therefore investigated whether a negative MAA outcome could have been predicted based on the evidence generated in the quality/CMC packages, regardless of clinical trial data. For this purpose, we reviewed the respective European Public Assessment Reports (EPARs) or withdrawal assessment reports, and the first regulatory assessments for all these 36 MAAs (i.e. day 120 of the centralized procedure), which are not publicly available. During EMA review, where significant issues are identified which would preclude a marketing authorisation, these issues are raised as questions to the applicant and are classified as major objections (MO).ResultsIn 67% of cases, the outcome of the quality and clinical assessment was the same, i.e. both the quality and clinical assessments either supported approval or did not support approval. In 11% of cases, MO were identified in the quality part of the submission but not in the clinical data. In 22% of cases, MO were raised on the clinical data package but not on the quality data. However, we found no instance where seemingly negative clinical data, including failed efficacy trials, led to a negative overall decision. In each instance, the failure to confirm similar clinical performance in all investigated aspects was eventually viewed as not being related to the biosimilar per se but as being due to imbalances in the trial arms, immaturity of secondary endpoint results, change in the reference product, or even chance findings. Furthermore, when performing an in-depth analysis of the quality and clinical packages of trastuzumab and rituximab biosimilars, we found that in no case were clinical trial data necessary to resolve residual uncertainties regarding the quality part.ConclusionThe results further support the argument that sufficient evidence for biosimilarity can be obtained from a combination of analytical and functional testing and pharmacokinetic studies which may also generate immunogenicity data. This calls into question the usefulness of comparative efficacy studies for the purposes of regulatory decision-making when approving biosimilar mAbs and fusion proteins.

Project description:ContextPatient-reported outcomes (PRO) data from randomised controlled trials (RCTs) are increasingly used to inform patient-centred care as well as clinical and health policy decisions.ObjectiveThe main objective of this study was to investigate the methodological quality of PRO assessment in RCTs of prostate cancer (PCa) and to estimate the likely impact of these studies on clinical decision making.Evidence acquisitionA systematic literature search of studies was undertaken on main electronic databases to retrieve articles published between January 2004 and March 2012. RCTs were evaluated on a predetermined extraction form, including (1) basic trial demographics and clinical and PRO characteristics; (2) level of PRO reporting based on the recently published recommendations by the International Society for Quality of Life Research; and (3) bias, assessed using the Cochrane Risk of Bias tool. Studies were systematically analysed to evaluate their relevance for supporting clinical decision making.Evidence synthesisSixty-five RCTs enrolling a total of 22 071 patients were evaluated, with 31 (48%) in patients with nonmetastatic disease. When a PRO difference between treatments was found, it related in most cases to symptoms only (n=29, 58%). Although the extent of missing data was generally documented (72% of RCTs), few reported details on statistical handling of this data (18%) and reasons for dropout (35%). Improvements in key methodological aspects over time were found. Thirteen (20%) RCTs were judged as likely to be robust in informing clinical decision making. Higher-quality PRO studies were generally associated with those RCTs that had higher internal validity.ConclusionsIncluding PRO in RCTs of PCa patients is critical for better evaluating the treatment effectiveness of new therapeutic approaches. Marked improvements in PRO quality reporting over time were found, and it is estimated that at least one-fifth of PRO RCTs have provided sufficient details to allow health policy makers and physicians to make critical appraisals of results.Patient summaryIn this report, we have investigated the methodological quality of PCa trials that have included a PRO assessment. We conclude that including PRO is critical to better evaluating the treatment effectiveness of new therapeutic approaches from the patient's perspective. Also, at least one-fifth of PRO RCTs in PCa have provided sufficient details to allow health policy makers and physicians to make a critical appraisal of results.

Dataset Information

Decision-making with multiple correlated binary outcomes in clinical trials.

Publications

Decision-making with multiple correlated binary outcomes in clinical trials.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets