Article type
Abstract
Background: The need for the efficient evaluation of the comparative effectiveness of medical interventions has led to the emergence of adaptative platform trials (APTs)—trials that simultaneously compare several interventions for a single condition with the option to add or remove interventions based on emerging evidence. As a result, systematic reviewers are increasingly faced with the challenge of integrating results from APTs in systematic reviews and meta-analyses.
Methods and objectives: We performed six systematic reviews that included evidence from APTs. These reviews addressed the effectiveness of therapeutics for acute COVID-19, community acquired pneumonia, and long COVID. We also performed a scoping review addressing methodological features of APTs.
Drawing from our experiences, we present unique considerations related to including evidence from APTs in systematic reviews and meta-analyses.
Results: We identified three key considerations. First, APTs’ flexibility to add or remove interventions may lead to comparisons against non-contemporaneous control groups. For example, a single control group, initiated at the inception of a trial, may be compared with several experimental interventions, each initiated at various points over the course of a trial. This may contribute to baseline imbalances between trial arms. Furthermore, APTs that occur over a prolonged period may also encounter drift in standard therapy or patient characteristics. Systematic reviewers will need to consider these issues in their assessment of risk of bias.
Second, the occurrence of overlapping control groups in APTs demands special attention in network meta-analyses—meta-analyses that compare three or more interventions simultaneously— to account for correlations between non-independent effect estimates from a single APT.
Finally, unlike conventional trials that rely on predefined sample sizes, most APTs use probabilities of futility, harm, or benefit to inform dynamic decisions about stopping interventions. Given that empirical evidence suggests that trials stopped early for benefit may inflate treatment effects, systematic reviewers must approach APTs with caution to avoid overly optimistic conclusions about treatment benefits.
Conclusions: APTs represent a promising approach to efficiently study the effectiveness of interventions. Their inclusion in systematic reviews and meta-analyses, however, warrants unique considerations, such as overlapping control groups, baseline imbalances between arms, and potential for overestimation of treatment effects.
Methods and objectives: We performed six systematic reviews that included evidence from APTs. These reviews addressed the effectiveness of therapeutics for acute COVID-19, community acquired pneumonia, and long COVID. We also performed a scoping review addressing methodological features of APTs.
Drawing from our experiences, we present unique considerations related to including evidence from APTs in systematic reviews and meta-analyses.
Results: We identified three key considerations. First, APTs’ flexibility to add or remove interventions may lead to comparisons against non-contemporaneous control groups. For example, a single control group, initiated at the inception of a trial, may be compared with several experimental interventions, each initiated at various points over the course of a trial. This may contribute to baseline imbalances between trial arms. Furthermore, APTs that occur over a prolonged period may also encounter drift in standard therapy or patient characteristics. Systematic reviewers will need to consider these issues in their assessment of risk of bias.
Second, the occurrence of overlapping control groups in APTs demands special attention in network meta-analyses—meta-analyses that compare three or more interventions simultaneously— to account for correlations between non-independent effect estimates from a single APT.
Finally, unlike conventional trials that rely on predefined sample sizes, most APTs use probabilities of futility, harm, or benefit to inform dynamic decisions about stopping interventions. Given that empirical evidence suggests that trials stopped early for benefit may inflate treatment effects, systematic reviewers must approach APTs with caution to avoid overly optimistic conclusions about treatment benefits.
Conclusions: APTs represent a promising approach to efficiently study the effectiveness of interventions. Their inclusion in systematic reviews and meta-analyses, however, warrants unique considerations, such as overlapping control groups, baseline imbalances between arms, and potential for overestimation of treatment effects.