Multi-arm multi-stage trials

A Multi-Arm Multi-Stage (MAMS) trial is an adaptive trial design used to evaluate multiple interventions within the same trial while allowing for the early stopping of ineffective treatment arms. This approach is particularly useful in fields such as oncology, chronic disease management, and public health, where testing multiple therapies efficiently is critical. MAMS designs improve resource use by combining the benefits of multi-arm comparison with planned interim analyses.

Defining the Research Question and Objectives

The primary objective of a MAMS trial is to identify the most effective intervention(s) among multiple options while stopping arms that show no evidence of benefit early. These trials are structured to test several treatments against a common control, using interim analyses to discontinue unpromising interventions. For example, a MAMS trial might ask: "Which chemotherapy regimen—Regimen A, B, or C—improves survival in patients with lung cancer?"

Key features of MAMS trials include:

Simultaneous testing of multiple interventions.
Predefined stages with interim analyses to continue or discontinue arms based on accumulating data.

Study Design Structure

MAMS trials are built on two design principles:

Multi-Arm: Multiple interventions are compared against a common control group within a single trial.
Multi-Stage: The trial is divided into distinct stages. At each stage, interim analyses evaluate the data to decide whether each arm should continue, stop for futility, or proceed to the next phase of evaluation.

For instance, a trial might include four arms: Control, Intervention A, Intervention B, and Intervention C, with interim analyses conducted every six months to assess effectiveness and safety.

Outcomes and Endpoints

The primary outcome in a MAMS trial is often a clinical measure such as survival rate, change in blood pressure, or event-free survival. Secondary outcomes may include safety, quality of life, treatment adherence, or cost-effectiveness. For example, in an oncology trial, the primary outcome might be 12-month survival, while secondary outcomes could include tumor response rates and patient-reported outcomes.

Interim Analysis Plan

Interim analyses are central to MAMS designs and are typically scheduled at predefined timepoints (e.g., every 6 months or after every 250 participants). These analyses assess whether individual treatment arms meet stopping criteria for futility, efficacy, or safety.

Stopping rules may include:

Futility: Stop an arm if it has less than a 20% probability of demonstrating benefit.
Efficacy: Continue or confirm an arm if the probability of success exceeds 80%.

Statistical methods commonly used include group sequential methods, Bayesian adaptive methods, and error-spending functions such as Lan-DeMets.

Sample Size Considerations

Sample size must be calculated with care, accounting for multiple comparisons and interim analyses. Adjustments such as the Bonferroni correction or alpha-spending approaches help maintain the family-wise error rate. For example, a MAMS trial with 1,000 participants might conduct interim analyses every 250 participants, with power calculations based on both the overall sample and planned stages.

Randomization and Allocation

Participants can be randomized to trial arms using equal or adaptive allocation. Response-adaptive randomization may be considered if early evidence suggests certain arms are more effective. This allows more participants to receive potentially beneficial treatments while preserving statistical integrity.

Implementation and Blinding

Blinding is ideal but not always feasible, especially for behavioral or surgical interventions. Drug trials may allow for full blinding. At minimum, outcome assessors should be blinded to reduce detection bias.

Data Collection and Monitoring

Consistent data collection across arms and timepoints is essential. MAMS trials benefit from centralized oversight by Data Monitoring Committees (DMCs), which evaluate interim results and make recommendations about continuing, stopping, or modifying treatment arms.

Statistical Analysis

The statistical plan should include:

Primary analysis: Comparing each intervention to the control using methods such as generalized linear models or survival analysis (e.g., Cox proportional hazards models).
Interim analyses: Guided by predefined rules for stopping arms based on performance.
Multiple comparisons: Appropriate adjustments to control Type I error across multiple arms.
Interaction analysis: Assessing whether effects differ among key subgroups.

Ethical Considerations

Ethical transparency is critical. Participants should be informed about the adaptive nature of the trial and the possibility of arms being stopped early. Safety monitoring must be robust, with DMC oversight. Trials should also ensure fair access to promising treatments, particularly as interim results emerge.

Reporting and Dissemination

MAMS trials should be reported following the CONSORT extension for adaptive trials. Reporting should include a clear account of how interim decisions were made, the results of each arm (regardless of whether they were stopped), and how the design contributed to trial efficiency.

Example: MAMS Trial Design

Research Question: Which intervention—Drug A, Drug B, or a lifestyle program—is most effective in reducing systolic blood pressure?

Design: A multi-arm (3 interventions + control), multi-stage trial with interim analyses at 6, 12, and 18 months.

Primary Outcome: Reduction in systolic blood pressure at 12 months.

Stopping Rules:

Stop arms with less than 20% chance of achieving a significant reduction.
Continue arms with greater than 80% probability of success.

Bibliography

Parmar MKB, Barthel FM-S, Sydes MR, et al. Speeding up the evaluation of new agents in cancer. Journal of the National Cancer Institute. 2008;100(17):1204–1214.
Sydes MR, Parmar MKB, Mason MD, et al. Flexible trial design in practice – stopping arms for lack-of-benefit and adding research arms mid-trial in STAMPEDE: a multi-arm multi-stage randomized controlled trial. Trials. 2012;13:168.
Jaki T. Multi-arm clinical trials with treatment selection: what can be gained and at what price? Clinical Investigation. 2015;5(4):293–301.
Royston P, Barthel FM-S, Parmar MKB. Designs for clinical trials with multiple primary outcomes: a practical approach. Statistical Methods in Medical Research. 2016;25(2):710–724.
Wason JMS, Jaki T, Stallard N. Planning multi-arm screening studies within the pharmaceutical industry. Statistics in Medicine. 2014;33(27):4639–4650.

Adapted for educational use. Please cite relevant trial methodology sources when using this material in research or teaching.