Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Improved up-and-down designs for phase I trials.

Anastasia Ivanova1, Aliakbar Montazer-Haghighi, Sri Gopal Mohanty

  • 1Department of Biostatistics, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7420, USA. aivanova@bios.unc.edu

Statistics in Medicine
|December 18, 2002
PubMed
Summary

This paper evaluates improved methods for determining safe drug dosages in early-stage clinical trials. By utilizing more patient data than traditional approaches, these new rules offer more precise and efficient ways to identify the target dose while minimizing toxicity risks.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Competing-triggering effect models for multitype recurrent event data.

Biometrics·2026
Same author

Dynamic borrowing methods for basket trials with order restrictions.

Journal of biopharmaceutical statistics·2026
Same author

PrecISE-a biomarker-stratified adaptive trial of 5 interventions in severe asthma: Final protocol and the baseline cohort.

The Journal of allergy and clinical immunology·2026
Same author

Design characteristics of sequential multiple assignment randomised trials (SMARTs) for human health: a scoping review of studies between 2009 and 2024.

BMJ open·2025
Same author

Prediction of transition probabilities in multi-state models with nested case-control data.

Biometrics·2025
Same author

Dynamic prediction by landmarking with data from cohort subsampling designs.

Statistical methods in medical research·2025

Area of Science:

  • Biostatistics and clinical trial methodology within up-and-down designs research
  • Pharmacology and drug development safety assessment

Background:

Early clinical trials often struggle to balance patient safety with the need for accurate dose identification. Traditional sequential allocation methods frequently rely solely on the most recent patient outcome to determine subsequent dosing levels. This limited perspective often leads to inefficient exploration of the dose-response curve. No prior work had fully resolved how to incorporate broader historical data into these simple allocation frameworks. Researchers have long sought ways to improve the precision of these trials without increasing sample sizes. That uncertainty drove the development of more sophisticated rules that look beyond the immediate preceding subject. These advancements aim to refine how investigators navigate the trade-off between toxicity and therapeutic efficacy. This study addresses these limitations by examining several enhanced strategies for sequential dose assignment.

Purpose Of The Study:

The aim of this study is to evaluate and improve sequential allocation rules for dose-response investigations. Researchers seek to overcome the limitations of traditional methods that rely solely on the most recent patient outcome. The study addresses the need for more efficient dose-finding strategies in early-stage clinical research. Investigators explore how incorporating historical data can enhance the precision of identifying the target dose. The motivation stems from the desire to minimize patient exposure to toxic levels while maximizing therapeutic potential. This work examines the performance of the k-in-a-row rule and introduces the novel Narayana rule. The authors also investigate which statistical estimators provide the most reliable results across varying sample sizes. By comparing these diverse approaches, the study provides a framework for optimizing the design of phase I trials.

Keywords:
dose-response studytoxicity probabilityisotonic regressionsequential allocation

Frequently Asked Questions

The Narayana rule improves dose allocation by utilizing a local estimate of toxicity probability derived from all previous patient responses. In contrast, traditional methods rely exclusively on the most recent outcome, which limits their ability to accurately identify the target dose level.

The k-in-a-row rule is a specific design that incorporates information from the k most recent patient responses. This approach contrasts with the Narayana rule, which aggregates all historical data to inform the assignment of the next dose level.

Isotonic regression is necessary for small to moderate sample sizes because it provides more accurate target dose estimates than alternative statistical estimators. The researchers propose that this method outperforms other approaches by better handling the inherent variability found in limited clinical datasets.

Related Experiment Videos

Main Methods:

Review approach involved evaluating multiple sequential allocation rules within the up-and-down family. The investigation focused on how different designs utilize patient response history to determine subsequent dosage levels. Researchers compared the performance of memoryless methods against those incorporating multiple past outcomes. The study design included the introduction of the Narayana rule, which relies on local toxicity probability estimates. The team assessed various statistical estimators to determine the most accurate way to identify the target dose. They performed simulations to observe how these rules behave as sample sizes grow. The approach prioritized identifying methods that minimize patient exposure to inappropriate dose levels. This systematic evaluation provided a comprehensive comparison of diverse allocation strategies and their associated estimation techniques.

Main Results:

Key findings from the literature indicate that utilizing more than the single most recent response improves the accuracy of dose-finding. The Narayana rule demonstrates that as sample sizes increase, assignments to dose levels distant from the target approach zero probability. This specific rule effectively focuses allocation on the two or three levels closest to the desired toxicity threshold. The analysis reveals that isotonic regression consistently outperforms other estimators when applied to small or moderate datasets. These results suggest that the choice of estimator is as critical as the allocation rule itself for trial success. The study confirms that incorporating historical data leads to more stable dose identification compared to traditional approaches. These findings provide empirical evidence that more complex rules yield better outcomes in early-stage research. The data show that these improvements are particularly beneficial for maintaining safety during the dose-escalation process.

Conclusions:

The authors demonstrate that incorporating historical data significantly enhances the performance of sequential dose allocation rules. Synthesis and implications suggest that the Narayana rule provides a robust framework for targeting specific toxicity probabilities. Their analysis indicates that utilizing broader information sets prevents unnecessary assignment to dose levels far from the target. The researchers propose that isotonic regression serves as a superior estimator for identifying the target dose in smaller studies. These findings imply that trial efficiency improves when investigators move away from memoryless allocation strategies. The study highlights the importance of selecting appropriate estimators to maximize the utility of collected patient responses. Their results provide a clear pathway for refining phase I trial protocols to better protect participants. Ultimately, these improved designs offer a more reliable approach to dose-finding in clinical research settings.

The Narayana rule uses all previous patient responses to calculate a local probability estimate. This data-driven approach ensures that as the sample size increases, the probability of assigning patients to dose levels far from the target effectively drops to zero.

The researchers measured the performance of different estimators by comparing their accuracy in identifying the target dose. They observed that isotonic regression consistently yielded superior results compared to other tested estimators when applied to small or moderate sample sizes.

The authors suggest that their findings support the adoption of more information-rich allocation rules in phase I trials. They propose that these methods enhance trial precision and safety, providing a more robust alternative to standard, less efficient sequential designs.