Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial

An example of a neonatal cooling trial

Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network, Pablo J. Sánchez

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Background: Decisions to stop randomized trials are often based on traditional P value thresholds and are often unconvincing to clinicians. To familiarize clinical investigators with the application and advantages of Bayesian monitoring methods, we illustrate the steps of Bayesian interim analysis using a recent major trial that was stopped based on frequentist analysis of safety and futility. Methods: We conducted Bayesian reanalysis of a factorial trial in newborn infants with hypoxic-ischemic encephalopathy that was designed to investigate whether outcomes would be improved by deeper (32 °C) or longer cooling (120 h), as compared with those achieved by standard whole body cooling (33.5 °C for 72 h). Using prior trial data, we developed neutral and enthusiastic prior probabilities for the effect on predischarge mortality, defined stopping guidelines for a clinically meaningful effect, and derived posterior probabilities for predischarge mortality. Results: Bayesian relative risk estimates for predischarge mortality were closer to 1.0 than were frequentist estimates. Posterior probabilities suggested increased predischarge mortality (relative risk > 1.0) for the three intervention groups; two crossed the Bayesian futility threshold. Conclusions: Bayesian analysis incorporating previous trial results and different pre-existing opinions can help interpret accruing data and facilitate informed stopping decisions that are likely to be meaningful and convincing to clinicians, meta-analysts, and guideline developers. Trial registration: ClinicalTrials.gov NCT01192776. Registered on 31 August 2010.

Original languageEnglish (US)
Article number335
JournalTrials
Volume17
Issue number1
DOIs
StatePublished - Jul 22 2016

Fingerprint

Bayes Theorem
Clinical Trials
Medical Futility
Mortality
Guidelines
Brain Hypoxia-Ischemia
Research Personnel
Newborn Infant
Safety

Keywords

  • Bayesian methods
  • Factorial trial
  • Hypothermia
  • Phase III trial
  • Stopping rules
  • Trial monitoring

ASJC Scopus subject areas

  • Medicine (miscellaneous)
  • Pharmacology (medical)

Cite this

Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network, & Sánchez, P. J. (2016). Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial: An example of a neonatal cooling trial. Trials, 17(1), [335]. https://doi.org/10.1186/s13063-016-1480-4

Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial : An example of a neonatal cooling trial. / Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network; Sánchez, Pablo J.

In: Trials, Vol. 17, No. 1, 335, 22.07.2016.

Research output: Contribution to journalArticle

Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network & Sánchez, PJ 2016, 'Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial: An example of a neonatal cooling trial', Trials, vol. 17, no. 1, 335. https://doi.org/10.1186/s13063-016-1480-4
Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network, Sánchez PJ. Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial: An example of a neonatal cooling trial. Trials. 2016 Jul 22;17(1). 335. https://doi.org/10.1186/s13063-016-1480-4
Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network ; Sánchez, Pablo J. / Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial : An example of a neonatal cooling trial. In: Trials. 2016 ; Vol. 17, No. 1.
@article{c63d7a2849694746a5295c007aba162c,
title = "Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial: An example of a neonatal cooling trial",
abstract = "Background: Decisions to stop randomized trials are often based on traditional P value thresholds and are often unconvincing to clinicians. To familiarize clinical investigators with the application and advantages of Bayesian monitoring methods, we illustrate the steps of Bayesian interim analysis using a recent major trial that was stopped based on frequentist analysis of safety and futility. Methods: We conducted Bayesian reanalysis of a factorial trial in newborn infants with hypoxic-ischemic encephalopathy that was designed to investigate whether outcomes would be improved by deeper (32 °C) or longer cooling (120 h), as compared with those achieved by standard whole body cooling (33.5 °C for 72 h). Using prior trial data, we developed neutral and enthusiastic prior probabilities for the effect on predischarge mortality, defined stopping guidelines for a clinically meaningful effect, and derived posterior probabilities for predischarge mortality. Results: Bayesian relative risk estimates for predischarge mortality were closer to 1.0 than were frequentist estimates. Posterior probabilities suggested increased predischarge mortality (relative risk > 1.0) for the three intervention groups; two crossed the Bayesian futility threshold. Conclusions: Bayesian analysis incorporating previous trial results and different pre-existing opinions can help interpret accruing data and facilitate informed stopping decisions that are likely to be meaningful and convincing to clinicians, meta-analysts, and guideline developers. Trial registration: ClinicalTrials.gov NCT01192776. Registered on 31 August 2010.",
keywords = "Bayesian methods, Factorial trial, Hypothermia, Phase III trial, Stopping rules, Trial monitoring",
author = "{Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network} and Claudia Pedroza and Tyson, {Jon E.} and Abhik Das and Abbot Laptook and Bell, {Edward F.} and Seetha Shankaran and Martin Keszler and Hensman, {Angelita M.} and Elisa Vierira and Emilee Little and Birju Shah and Nicholas Guerina and Joseph Bliss and Hussnain Mirza and Ross Sommers and Walsh, {Michele C.} and Hibbs, {Anna Maria} and Newman, {Nancy S.} and Siner, {Bonnie S.} and Arlene Zadell and Truog, {William E.} and Pallotto, {Eugenia K.} and Kilbride, {Howard W.} and Cheri Gauldin and Anne Holmes and Kathy Johnson and Kurt Schibler and Kallapur, {Suhas G.} and Barbara Alexander and Fischer, {Estelle E.} and Gratton, {Teresa L.} and Cathy Grisby and Lenora Jackson and Jennifer Jennings and Kristin Kirker and Greg Muthig and Sandra Wuertz and {Michael Cotten}, C. and Goldberg, {Ronald N.} and Joanne Finkle and Fisher, {Kimberley A.} and Sandra Grimes and Laughon, {Matthew M.} and Bose, {Carl L.} and Janice Bernhardt and S{\'a}nchez, {Pablo J.} and Stoll, {Barbara J.} and Myra Wyckoff and Chalak, {Lina F.} and Brion, {Luc P.}",
year = "2016",
month = "7",
day = "22",
doi = "10.1186/s13063-016-1480-4",
language = "English (US)",
volume = "17",
journal = "Trials",
issn = "1745-6215",
publisher = "BioMed Central",
number = "1",

}

TY - JOUR

T1 - Advantages of Bayesian monitoring methods in deciding whether and when to stop a clinical trial

T2 - An example of a neonatal cooling trial

AU - Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network

AU - Pedroza, Claudia

AU - Tyson, Jon E.

AU - Das, Abhik

AU - Laptook, Abbot

AU - Bell, Edward F.

AU - Shankaran, Seetha

AU - Keszler, Martin

AU - Hensman, Angelita M.

AU - Vierira, Elisa

AU - Little, Emilee

AU - Shah, Birju

AU - Guerina, Nicholas

AU - Bliss, Joseph

AU - Mirza, Hussnain

AU - Sommers, Ross

AU - Walsh, Michele C.

AU - Hibbs, Anna Maria

AU - Newman, Nancy S.

AU - Siner, Bonnie S.

AU - Zadell, Arlene

AU - Truog, William E.

AU - Pallotto, Eugenia K.

AU - Kilbride, Howard W.

AU - Gauldin, Cheri

AU - Holmes, Anne

AU - Johnson, Kathy

AU - Schibler, Kurt

AU - Kallapur, Suhas G.

AU - Alexander, Barbara

AU - Fischer, Estelle E.

AU - Gratton, Teresa L.

AU - Grisby, Cathy

AU - Jackson, Lenora

AU - Jennings, Jennifer

AU - Kirker, Kristin

AU - Muthig, Greg

AU - Wuertz, Sandra

AU - Michael Cotten, C.

AU - Goldberg, Ronald N.

AU - Finkle, Joanne

AU - Fisher, Kimberley A.

AU - Grimes, Sandra

AU - Laughon, Matthew M.

AU - Bose, Carl L.

AU - Bernhardt, Janice

AU - Sánchez, Pablo J.

AU - Stoll, Barbara J.

AU - Wyckoff, Myra

AU - Chalak, Lina F.

AU - Brion, Luc P.

PY - 2016/7/22

Y1 - 2016/7/22

N2 - Background: Decisions to stop randomized trials are often based on traditional P value thresholds and are often unconvincing to clinicians. To familiarize clinical investigators with the application and advantages of Bayesian monitoring methods, we illustrate the steps of Bayesian interim analysis using a recent major trial that was stopped based on frequentist analysis of safety and futility. Methods: We conducted Bayesian reanalysis of a factorial trial in newborn infants with hypoxic-ischemic encephalopathy that was designed to investigate whether outcomes would be improved by deeper (32 °C) or longer cooling (120 h), as compared with those achieved by standard whole body cooling (33.5 °C for 72 h). Using prior trial data, we developed neutral and enthusiastic prior probabilities for the effect on predischarge mortality, defined stopping guidelines for a clinically meaningful effect, and derived posterior probabilities for predischarge mortality. Results: Bayesian relative risk estimates for predischarge mortality were closer to 1.0 than were frequentist estimates. Posterior probabilities suggested increased predischarge mortality (relative risk > 1.0) for the three intervention groups; two crossed the Bayesian futility threshold. Conclusions: Bayesian analysis incorporating previous trial results and different pre-existing opinions can help interpret accruing data and facilitate informed stopping decisions that are likely to be meaningful and convincing to clinicians, meta-analysts, and guideline developers. Trial registration: ClinicalTrials.gov NCT01192776. Registered on 31 August 2010.

AB - Background: Decisions to stop randomized trials are often based on traditional P value thresholds and are often unconvincing to clinicians. To familiarize clinical investigators with the application and advantages of Bayesian monitoring methods, we illustrate the steps of Bayesian interim analysis using a recent major trial that was stopped based on frequentist analysis of safety and futility. Methods: We conducted Bayesian reanalysis of a factorial trial in newborn infants with hypoxic-ischemic encephalopathy that was designed to investigate whether outcomes would be improved by deeper (32 °C) or longer cooling (120 h), as compared with those achieved by standard whole body cooling (33.5 °C for 72 h). Using prior trial data, we developed neutral and enthusiastic prior probabilities for the effect on predischarge mortality, defined stopping guidelines for a clinically meaningful effect, and derived posterior probabilities for predischarge mortality. Results: Bayesian relative risk estimates for predischarge mortality were closer to 1.0 than were frequentist estimates. Posterior probabilities suggested increased predischarge mortality (relative risk > 1.0) for the three intervention groups; two crossed the Bayesian futility threshold. Conclusions: Bayesian analysis incorporating previous trial results and different pre-existing opinions can help interpret accruing data and facilitate informed stopping decisions that are likely to be meaningful and convincing to clinicians, meta-analysts, and guideline developers. Trial registration: ClinicalTrials.gov NCT01192776. Registered on 31 August 2010.

KW - Bayesian methods

KW - Factorial trial

KW - Hypothermia

KW - Phase III trial

KW - Stopping rules

KW - Trial monitoring

UR - http://www.scopus.com/inward/record.url?scp=84979021974&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84979021974&partnerID=8YFLogxK

U2 - 10.1186/s13063-016-1480-4

DO - 10.1186/s13063-016-1480-4

M3 - Article

VL - 17

JO - Trials

JF - Trials

SN - 1745-6215

IS - 1

M1 - 335

ER -