Table 2 Innovations in randomized controlled trials and observational studies

Adaptive trials

Adaptive methods include scheduled interim looks at the data during the trial. This leads to predetermined changes based on the analyses of accumulating data, all the while maintaining trial validity and integrity [15]

Despite being more complex than traditional RCTs, adaptive trials can bring about numerous benefits, such as shortening trial duration or obtaining more precise conclusions [15]

Jardine et al. 2022 [25]

Wang et al. 2018 [26]

Sequential trials

Sequential trials are an approach to clinical trials during which subjects are serially recruited and study results are continuously analyzed [16]. Once enough data enabling a decision regarding treatment effectiveness is collected, the trial is stopped [17]

Since a sequential trial can be halted as soon as treatment efficacy or lack thereof is demonstrated, a reliable result is obtained with a minimum number of patients [16]

Lewis and Bessen 1990 [16]

Gu et al. 2015 [27]

Platform trials

Platform trials are a type of clinical trial during which multiple interventions can be compared simultaneously to a common control group within a single master protocol [28]

With a platform trial, having a common control arm can decrease the number of patients to be enrolled, the cost, and the time of a RCT [29]

Parker et al. 2018 [30]

Yee et al. 2022 [31]

EHR (Electronic Health Record)-based clinical trials

EHRs and an expanded access to routinely-collected clinical data has resulted in RCTs being conducted within the context of EHR-based clinical trials. [19]

EHRs may facilitate pre-screening of patients by age, sex, and diagnosis, helping to exclude ineligible patients, and reduce the overall screening duration in clinical trials [32]

Price et al. 2017 [33]

Bereznicki et al. 2008 [34]

Observational studies

Innovation

Definition

Strengths

Application of the method

Causal inference methods

Causal inference in observational studies refers to an intellectual discipline which allows researchers to draw causal conclusions based on data by considering the assumptions, study design, and estimation strategies [20]

Causal inference methods, through their well-defined frameworks and assumptions have helped to overcome concerns about bias in the analysis of observational studies [10]

Ekline et al. 2011[35]

Skerritt et al. 2021 [36]

DAG (Directed acyclic graph)

When considering the effect of one variable on another, DAGs serve as a visual representation of causal assumptions. This structured approach moves the conversation forward by serving as a visual aid that makes underlying relations explicit [37]

DAGs can help identify possible confounding for the causal question being considered [37]

Pakzad et al. 2023 [38]

Byrne et al. 2019 [39]

E-value

The E-value is “the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the treatment and the outcome to fully explain away a specific treatment-outcome association, conditional on the measured covariates” [24]

The E-value is an intuitive metric to help determine how robust the results of a study are to unmeasured confounding [24]

Bender Ignacio et al. 2018 [40]

Eastwood et al. 2018 [41]

Use of “big data”

Large observational studies have become more popular in the era of big data because of their ability to leverage and analyze multiple sources of observational data [22] such as from population databases, social media, and digital health tools [23]

Use of big data in research can help with hypothesis generating, and focuses on the temporal stability of the association [23]

Khera et al. 2018 [42]

Ahmed et al. 2023 [43]