TREATMENT SWITCHING: STATISTICAL AND DECISION-MAKING CHALLENGES AND APPROACHES

Area of Science:

• Biostatistics
• Clinical Trials
• Health Economics

Background:

• Treatment switching, where participants change treatments post-randomization in clinical trials, challenges standard intention-to-treat analyses.
• Estimating true comparative effectiveness is difficult when control group patients switch to experimental treatments.

Purpose of the Study:

• To explain statistical methods for adjusting for treatment switching in clinical trials for a non-statistician audience.
• To summarize stakeholder perspectives on the acceptability and utility of these adjustment methods.

Main Methods:

• Description of three statistical adjustment methods: marginal structural models, two-stage adjustment, and rank preserving structural failure time models.
• Exploration of stakeholder views on method acceptability, gathered from the 2014 Adelaide International Workshop.

Main Results:

• Stakeholders acknowledged that adjustment methods rely on potentially questionable assumptions.
• Disagreement exists regarding the acceptability of adjustment methods, with consensus on the need for rigorous justification.
• The utility of adjustment methods is context-dependent, varying with the decision-making process.

Conclusions:

• Treatment switching poses significant challenges for estimating treatment comparative effects.
• Decision-makers express reservations about adjustment methods, necessitating further investigation into their utility and acceptability.
• Development of adjustment methods needs to address real-world needs and improve decision-maker acceptance.