Urn models for response-adaptive randomized designs

a simulation study based on a non-adaptive randomized trial

Andrea Ghiglietti, Maria Giovanna Scarale, Rosalba Miceli, Francesca Ieva, Luigi Mariani, Cecilia Gavazzi, Anna Maria Paganoni, Valeria Edefonti

Research output: Contribution to journalArticle

Abstract

Recently, response-adaptive designs have been proposed in randomized clinical trials to achieve ethical and/or cost advantages by using sequential accrual information collected during the trial to dynamically update the probabilities of treatment assignments. In this context, urn models-where the probability to assign patients to treatments is interpreted as the proportion of balls of different colors available in a virtual urn-have been used as response-adaptive randomization rules. We propose the use of Randomly Reinforced Urn (RRU) models in a simulation study based on a published randomized clinical trial on the efficacy of home enteral nutrition in cancer patients after major gastrointestinal surgery. We compare results with the RRU design with those previously published with the non-adaptive approach. We also provide a code written with the R software to implement the RRU design in practice. In detail, we simulate 10,000 trials based on the RRU model in three set-ups of different total sample sizes. We report information on the number of patients allocated to the inferior treatment and on the empirical power of the t-test for the treatment coefficient in the ANOVA model. We carry out a sensitivity analysis to assess the effect of different urn compositions. For each sample size, in approximately 75% of the simulation runs, the number of patients allocated to the inferior treatment by the RRU design is lower, as compared to the non-adaptive design. The empirical power of the t-test for the treatment effect is similar in the two designs.

Original languageEnglish
Pages (from-to)1203-1215
Number of pages13
JournalJournal of Biopharmaceutical Statistics
Volume28
Issue number6
DOIs
Publication statusPublished - 2018

Fingerprint

Urn model
Randomized Trial
Simulation Study
Randomized Clinical Trial
t-test
Sample Size
Therapeutics
Randomized Controlled Trials
Adaptive Design
Nutrition
Treatment Effects
Randomisation
Surgery
Enteral Nutrition
Sensitivity Analysis
Assign
Efficacy
Random Allocation
Cancer
Ball

Cite this

Urn models for response-adaptive randomized designs : a simulation study based on a non-adaptive randomized trial. / Ghiglietti, Andrea; Scarale, Maria Giovanna; Miceli, Rosalba; Ieva, Francesca; Mariani, Luigi; Gavazzi, Cecilia; Paganoni, Anna Maria; Edefonti, Valeria.

In: Journal of Biopharmaceutical Statistics, Vol. 28, No. 6, 2018, p. 1203-1215.

Research output: Contribution to journalArticle

@article{0924af3c8f814a8a8cb540a4e68b45fa,
title = "Urn models for response-adaptive randomized designs: a simulation study based on a non-adaptive randomized trial",
abstract = "Recently, response-adaptive designs have been proposed in randomized clinical trials to achieve ethical and/or cost advantages by using sequential accrual information collected during the trial to dynamically update the probabilities of treatment assignments. In this context, urn models-where the probability to assign patients to treatments is interpreted as the proportion of balls of different colors available in a virtual urn-have been used as response-adaptive randomization rules. We propose the use of Randomly Reinforced Urn (RRU) models in a simulation study based on a published randomized clinical trial on the efficacy of home enteral nutrition in cancer patients after major gastrointestinal surgery. We compare results with the RRU design with those previously published with the non-adaptive approach. We also provide a code written with the R software to implement the RRU design in practice. In detail, we simulate 10,000 trials based on the RRU model in three set-ups of different total sample sizes. We report information on the number of patients allocated to the inferior treatment and on the empirical power of the t-test for the treatment coefficient in the ANOVA model. We carry out a sensitivity analysis to assess the effect of different urn compositions. For each sample size, in approximately 75{\%} of the simulation runs, the number of patients allocated to the inferior treatment by the RRU design is lower, as compared to the non-adaptive design. The empirical power of the t-test for the treatment effect is similar in the two designs.",
author = "Andrea Ghiglietti and Scarale, {Maria Giovanna} and Rosalba Miceli and Francesca Ieva and Luigi Mariani and Cecilia Gavazzi and Paganoni, {Anna Maria} and Valeria Edefonti",
year = "2018",
doi = "10.1080/10543406.2018.1452024",
language = "English",
volume = "28",
pages = "1203--1215",
journal = "Journal of Biopharmaceutical Statistics",
issn = "1054-3406",
publisher = "Taylor and Francis Ltd.",
number = "6",

}

TY - JOUR

T1 - Urn models for response-adaptive randomized designs

T2 - a simulation study based on a non-adaptive randomized trial

AU - Ghiglietti, Andrea

AU - Scarale, Maria Giovanna

AU - Miceli, Rosalba

AU - Ieva, Francesca

AU - Mariani, Luigi

AU - Gavazzi, Cecilia

AU - Paganoni, Anna Maria

AU - Edefonti, Valeria

PY - 2018

Y1 - 2018

N2 - Recently, response-adaptive designs have been proposed in randomized clinical trials to achieve ethical and/or cost advantages by using sequential accrual information collected during the trial to dynamically update the probabilities of treatment assignments. In this context, urn models-where the probability to assign patients to treatments is interpreted as the proportion of balls of different colors available in a virtual urn-have been used as response-adaptive randomization rules. We propose the use of Randomly Reinforced Urn (RRU) models in a simulation study based on a published randomized clinical trial on the efficacy of home enteral nutrition in cancer patients after major gastrointestinal surgery. We compare results with the RRU design with those previously published with the non-adaptive approach. We also provide a code written with the R software to implement the RRU design in practice. In detail, we simulate 10,000 trials based on the RRU model in three set-ups of different total sample sizes. We report information on the number of patients allocated to the inferior treatment and on the empirical power of the t-test for the treatment coefficient in the ANOVA model. We carry out a sensitivity analysis to assess the effect of different urn compositions. For each sample size, in approximately 75% of the simulation runs, the number of patients allocated to the inferior treatment by the RRU design is lower, as compared to the non-adaptive design. The empirical power of the t-test for the treatment effect is similar in the two designs.

AB - Recently, response-adaptive designs have been proposed in randomized clinical trials to achieve ethical and/or cost advantages by using sequential accrual information collected during the trial to dynamically update the probabilities of treatment assignments. In this context, urn models-where the probability to assign patients to treatments is interpreted as the proportion of balls of different colors available in a virtual urn-have been used as response-adaptive randomization rules. We propose the use of Randomly Reinforced Urn (RRU) models in a simulation study based on a published randomized clinical trial on the efficacy of home enteral nutrition in cancer patients after major gastrointestinal surgery. We compare results with the RRU design with those previously published with the non-adaptive approach. We also provide a code written with the R software to implement the RRU design in practice. In detail, we simulate 10,000 trials based on the RRU model in three set-ups of different total sample sizes. We report information on the number of patients allocated to the inferior treatment and on the empirical power of the t-test for the treatment coefficient in the ANOVA model. We carry out a sensitivity analysis to assess the effect of different urn compositions. For each sample size, in approximately 75% of the simulation runs, the number of patients allocated to the inferior treatment by the RRU design is lower, as compared to the non-adaptive design. The empirical power of the t-test for the treatment effect is similar in the two designs.

U2 - 10.1080/10543406.2018.1452024

DO - 10.1080/10543406.2018.1452024

M3 - Article

VL - 28

SP - 1203

EP - 1215

JO - Journal of Biopharmaceutical Statistics

JF - Journal of Biopharmaceutical Statistics

SN - 1054-3406

IS - 6

ER -