Clinical Trials

Response-adaptive randomization is controversial even in the best circumstances when based on a quickly determined primary outcome. In disease settings in which the primary outcome requires long follow-up, an intermediate endpoint may be chosen to update randomization allocations. The aim of our study is to evaluate the impact of response-adaptive randomization applied to an imperfect intermediate endpoint. We use tuberculosis trials as the motivating example.

We simulated a response-adaptive randomization design, adapting randomization allocations using an imperfect intermediate endpoint, in a superiority trial of two experimental regimens and one control arm. The primary study outcome was treatment success after 73 weeks from randomization; the intermediate endpoint was culture conversion at 8 weeks. We compared different sensitivity (Se) and specificity (Spe) scenarios for the intermediate endpoint, while varying the true treatment efficacy. We evaluated the performance of response-adaptive randomization to achieve its primary goal of allocating more participants to the better arm and the impact of time-trends on type I error rate.

Even in an ideal state of perfect accuracy (i.e. intermediate endpoint with Se = 100% and Spe = 100%), response-adaptive randomization did not always live up to its main purpose of allocating more patients to the better arm. Lower accuracy of the intermediate endpoint leads to greater divergence from the goal of more allocations to the better arm. The larger the difference in treatment efficacy between the arms, the more striking the impact of an intermediate endpoint with poor diagnostic accuracy. Time-trends inflate the type I error rate, and while stratified tests can correct this, they do so at the cost of a power loss. Allocating more patients to the worst arm increases power for comparisons with this arm but reduces power for comparisons of the best arm to control.

Given the objective of evaluating several new therapeutic regimens in a timely manner, response-adaptive randomization is tempting. However, it requires at least reliance on highly accurate intermediate endpoints, which are still no guarantee of response-adaptive randomization’s trustworthiness.

Cluster randomised crossover designs (CRXOs) are a powerful type of longitudinal cluster randomised trial in which all participating clusters switch between two treatment conditions. “Multiple-period” CRXOs divide the trial duration into a number of periods of equal length and can allow for multiple crossovers between treatment conditions. It can be assumed that increasing the number of crossovers leads to an increase in statistical power. We investigate whether this is true for standard correlation structures, comparing CRXO designs with equal numbers of clusters, participants, and periods but differing in the number of crossovers, considering continuous outcomes.

We consider the formula for the variance of the treatment effect estimator for multiple-period CRXOs under exchangeable and block-exchangeable within-cluster correlation structures, assuming equal cluster-period sizes and different patterns of treatment conditions in the treatment sequences varying in the number of crossovers. We also conduct a simulation study to compare the statistical power between multiple-period CRXO designs with different numbers of crossovers that share the same number and duration of periods and the same number of participants in each cluster period.

Under exchangeable and block-exchangeable correlation structures and equal cluster-period sizes, the number of crossovers in the treatment sequences does not impact study power, provided the design is balanced in terms of the number of periods and clusters implementing each condition.

When an exchangeable or block-exchangeable within-cluster correlation structure and a time-invariant effect of treatment are assumed, power calculations for CRXO designs are invariant to the specific ordering of treatment conditions. In particular, a CRXO design with additional crossovers does not lead to increased statistical power compared to a CRXO design with just one crossover for these within-cluster correlation structures. Further work is required to investigate the utility of multiple crossovers in situations where the treatment effect varies over time.

There is growing interest in leveraging real-world data, such as electronic health records, administrative claims data, and patient registries, to generate real-world evidence studies that support the U.S. Food and Drug Administration’s premarket and postmarket regulatory determinations of effectiveness and/or safety for novel therapeutics. We examined the frequency and characteristics of real-world evidence studies used by the U.S. Food and Drug Administration to support premarket determinations of effectiveness and/or safety, as well as those required or requested by the U.S. Food and Drug Administration to be conducted postmarket after approval.

We identified all novel therapeutics approved by the U.S. Food and Drug Administration between 2016 and 2024, using action packages from the Drugs@FDA database. Product labels, approval letters, and review documents were used to identify real-world evidence studies supporting premarket determinations of effectiveness and/or safety, as well as all postmarketing requirements or commitments outlined at the time of approval. Outcomes included the number of novel therapeutics approved with premarket and/or postmarket real-world evidence studies and characteristics of these studies, including study design, data source, and primary objectives.

From 2016 to 2024, the U.S. Food and Drug Administration approved 400 novel therapeutics for 543 indications, of which 43 (10.8%) had at least one real-world evidence study that supported premarket determinations of effectiveness and/or safety (64 unique studies), and 138 (34.5%) had at least one real-world evidence study required or requested by the U.S. Food and Drug Administration to be conducted postmarket after approval (208 unique studies). Among the 64 unique premarket real-world evidence studies, the most common study designs were non-interventional (observational) studies (35, 54.7%) and externally controlled trials (17, 26.6%); 38 (59.4%) studies utilized electronic health or medical records, and 47 (73.4%) provided evidence on effectiveness. Among the 208 unique postmarket real-world evidence studies, the most common study design was non-interventional (observational) studies (159, 76.4%); 61 (29.3%) studies identified registries as the proposed data source, and 197 (94.7%) were designed to provide evidence on safety alone. The proportion of therapeutics approved with at least one postmarket real-world evidence study increased over time from 2 of 20 (10.0%) in 2016 to 23 of 47 (48.9%) in 2024; however, only 7 (3.4%) of these studies were classified by the U.S. Food and Drug Administration as fulfilled or submitted as of May 2025.

Real-world evidence studies are infrequently used to support the U.S. Food and Drug Administration’s premarket determinations of effectiveness and/or safety but have been increasingly required or requested by the U.S. Food and Drug Administration to be conducted postmarket after approval; however delays in completing postmarket real-world evidence studies may limit their regulatory impact.

Stepped wedge designs are a specific type of cluster randomised trial design that requires all clusters to start and end the study at the same time. Batched stepped wedge designs are a variant that allows clusters to commence the study in batches instead of all at once. Since different batches of clusters may have different characteristics, it is possible that the treatment effect differs across batches. Here, we explore different approaches for accommodating this treatment effect heterogeneity when analysing data from a batched stepped wedge design where each participant provides a single outcome measurement: linear mixed models with a random coefficient for the intervention for each batch of clusters with estimation via restricted maximum likelihood; and meta-analysis approaches that pool the batch-specific treatment effect estimates using fixed- or random-effects meta-analysis approaches. In many batched stepped wedge designs, the number of batches will be low (e.g. 2); it is unclear what impact the small number of batches may have on bias and confidence interval coverage of these approaches.

We simulated continuous outcome data from a range of batched stepped wedge designs with two and five batches, different within-cluster correlation structures (exchangeable, block-exchangeable), with and without period effects shared across batches, and with and without treatment effect heterogeneity across batches. Using the ‘lme4’ and ‘meta’ packages in R 4.4.1, we analysed each simulated data set using both linear mixed effects models (with and without treatment effect heterogeneity) and a range of meta-analysis approaches. We assessed the relative performance of all methods to estimate the true treatment effect using rates of convergence, bias, confidence interval coverage, and empirical standard error.

Treatment effect estimates from all meta-analysis approaches were unbiased. When period effects were incorrectly specified, treatment effects from linear mixed effects models were biased; when correctly specified, treatment effect estimates from these models were unbiased. Confidence interval coverage was similar for all random-effects meta-analysis approaches and close to the nominal level. For fixed-effect meta-analysis and linear mixed models, coverage depended on whether treatment effect heterogeneity was present.

When batched stepped wedge trials include a small number of batches (two and five batches), linear mixed models provide unbiased estimates of treatment effects when correctly specified, but exhibit under-coverage of confidence intervals when treatment effect heterogeneity is present. Random-effects meta-analysis approaches provide unbiased estimates and nominal confidence interval coverage whether or not period or treatment effects vary across batches, but can yield excessively wide confidence intervals. When batched stepped wedge trials include small numbers of batches, we recommend the use of random-effects meta-analysis approaches for the analysis of continuous outcomes.

Multiple imputation is often recommended over complete case analysis for handling missing data in clinical trials due to its ability to recover information from participants with incomplete data. While multiple imputation is generally held to be more efficient than complete case analysis, its planned use in clinical trials is typically not considered during sample size estimation. The standard approach of inflating the sample size for anticipated loss to follow-up is applicable for complete case analysis but could lead to excess power and hence inefficient resource use should multiple imputation be planned for analysis. In this article, we systematically reviewed published clinical trials with the aim of quantifying the precision advantages of multiple imputation over complete case analysis in treatment effect estimation, hence informing sample size planning for future trials.

We conducted a targeted review of clinical trials published between January 2019 and December 2023 in Lancet, The BMJ, Journal of the American Medical Association and New England Journal of Medicine. Clinical trials were eligible for inclusion if point and variance estimates for the effect of treatment on a primary efficacy or safety outcome could be determined for both multiple imputation and complete case analysis. The design effect due to multiple imputation was calculated as the variance of the treatment effect estimate using multiple imputation divided by the corresponding variance using complete case analysis. As a supplementary analysis, we also conducted an untargeted review of other journals by searching in PubMed for clinical trials with the keywords ‘imputation’ or ‘imputed’ in their title or abstract.

The targeted search identified 547 articles, of which 59 satisfied eligibility criteria. Included trials tended to be large in size (median 653 participants) and reported a median of 8.6% missing data in the complete case analysis of the primary outcome (range 0.4%–30.5%). Multiple imputation was most frequently applied using chained equations under a missing at random assumption, with auxiliary variables included in the imputation model in most trials. The median design effect due to multiple imputation was 1.00 in both unadjusted (n = 15 trials) and covariate-adjusted analyses (n = 46 trials), suggesting multiple imputation typically was not offering precision advantages over complete case analysis. Similar design effects were observed in the untargeted review (median 0.96 and 1.01 for unadjusted and covariate-adjusted analyses), despite higher rates of missing data overall (median 15.7%, n = 49 trials).

Multiple imputation did not consistently lead to more precise treatment effect estimates than complete case analysis in the trials included in the review. Findings should not be construed as an argument against the use of multiple imputation but suggest the standard approach of inflating the sample size for anticipated loss to follow-up is reasonable when multiple imputation is planned for analysis.

Statistical analysis plans are critical regulatory documents that define the statistical methodology, objectives, and data-handling procedures for clinical trials. The traditional process of statistical analysis plan development is resource-intensive, typically spanning 4–6 weeks, and is increasingly complicated by evolving regulatory requirements and complex trial designs such as adaptive studies and real-world evidence investigations. The emergence of generative artificial intelligence presents an opportunity to streamline this workflow; however, applying AI in regulated environments requires rigorous control over accuracy and consistency.

To address these challenges, we designed and implemented a generative artificial intelligence–based solution tailored to the automated drafting of statistical analysis plans. The system utilizes a knowledge graph and vector database architecture to parse clinical protocols and map content to study-specific templates. We employed a hybrid generation strategy featuring four distinct modes: verbatim copying, summarization, standard text with dynamic variable insertion, and de novo text generation via prompt engineering. The solution was deployed as both a web application and a Microsoft Word add-in and was evaluated across a diverse portfolio of trial designs, including interventional, non-interventional, clinical pharmacology, and oncology studies.

The deployment of the solution yielded a transformative reduction in drafting time, creating first drafts in an average of 1.0–3.4 min across 71 statistical analysis plans, compared to the traditional timeline of 1–2 days. Subject matter expert evaluations rated the outputs between 3.6 and 4.2 on a five-point Likert-type scale, indicating moderate to high quality. Furthermore, in comparisons against manually authored statistical analysis plans, a Large Language Model–assisted evaluation demonstrated substantial preservation of meaning across corresponding sections, yielding an average semantic similarity score of 0.75. Automated standardization features ensured a high degree of consistency across diverse statistical analysis plan documents. This approach enables the management of expansive statistical analysis plan portfolios and supports a broad spectrum of trial designs, including adaptive studies and real-world evidence investigations.

The application of generative artificial intelligence to statistical analysis plan authoring represents a substantive advancement for clinical research operations, delivering marked improvements in drafting speed and documentation consistency. Beyond immediate efficiency gains, standardized automated text generation may benefit downstream processes such as statistical programming and clinical study report drafting. Future work should focus on strengthening automated quality assurance, supporting amendments with auditable revision handling, and further integrating artificial intelligence–assisted authoring into end-to-end clinical trial documentation workflows.

In this article, we introduce a Phase I clinical trial design that utilizes late-onset toxicity outcomes to determine group-specific doses when groups are partially ordered before the trial. By “partially ordered groups,” we mean that the frailty order is known between some, but not all, groups. Previous research has addressed dose finding for partially ordered groups and for late-onset toxicity separately, but no solution has been proposed for dose finding that incorporates both aspects simultaneously.

Motivated by a Phase Ib study involving heterogeneous groups with metastatic castration-resistant prostate cancer, we extend the continual reassessment method shift model framework to tackle the problem of dose finding in partially ordered groups with late-onset toxicity. Our simulations compare the proposed method with an approach that incorporates group information without considering late-onset toxicity. An alternative strategy could involve conducting independent parallel trials for each group; however, this approach risks dose reversals, where a more sensitive group is assigned a higher dose than a less sensitive group.

In comparison with a similar dose-finding method that does not incorporate late-onset toxicity, our proposed method shows similar performance in selecting the correct dose. Through an extensive simulation study, the generated results demonstrate good operating characteristics for the proposed method in terms of the proportion of correct dose selection within groups, while accounting for late-onset toxicities.

Our proposed method avoids dose reversals and demonstrates favorable operating characteristics in recommending appropriate doses for each group.

The Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results trial randomized patients with diabetes to liraglutide or placebo. The per-protocol analysis conditioned on post-baseline adherence and therefore lacked a well-defined estimand. This study sought to evaluate a subset of Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results to estimate the effect of sustained liraglutide exposure corresponding to a well-defined per-protocol analysis.

We used the roadmap of targeting learning to define the estimand for the sustained treatment analysis and estimated it with longitudinal targeted minimum loss-based estimation, accounting for protocol deviations and censoring using post-baseline confounders. The results were compared to the intention-to-treat analysis. We report 3.5-year risks of the primary composite outcome (myocardial infarction, stroke, or cardiovascular mortality) and secondary outcomes (each of the primary outcome’s components, revascularization, unstable angina pectoris, heart failure, and all-cause mortality).

The intention-to-treat analysis estimated 3.5-year risks for the primary composite outcome of 11.8% (95% confidence interval: 10.8 to 12.8) in the liraglutide arm and 13.3% (95% confidence interval: 12.3 to 14.3) in the placebo arm, with a risk difference of 1.5% (95% confidence interval: 0.1 to 2.9). Accounting for post-baseline confounders, the sustained treatment analysis estimated 3.5-year risks of 11.4% (95% confidence interval: 10.4 to 12.5) for sustained exposure to liraglutide versus 12.6% (95% confidence interval: 11.5 to 13.7) for sustained exposure to placebo, yielding a risk difference of 1.1% (95% confidence interval: −0.4 to 2.6). Sustained exposure to liraglutide showed no statistically significant difference compared with sustained exposure to placebo for any secondary outcomes at 3.5 years.

We used contemporary methods for causal inference to define and estimate the sustained treatment effects of the Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results study by accounting for both baseline and time-varying confounders. The results were consistent with the findings of the original Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results trial.

The strength and transparency of clinical trial evidence supporting drug approvals have become increasingly scrutinized, particularly considering the increased use of regulatory flexibility and expedited pathways. While US Food and Drug Administration standards have been extensively analyzed, evidence standards at the European Medicines Agency remain less well-characterized. Thus, this study aims to systematically assess the design, quality, and outcomes of pivotal efficacy trials supporting European Medicines Agency drug approvals for new active substances between 2020 and 2023.

We conducted a cross-sectional analysis of drugs receiving positive opinions from the European Medicines Agency’s Committee for Medicinal Products for Human Use and subsequent approval by the European Commission between January 2020 and December 2023. Data were extracted from European Public Assessment Reports and European Medicines Agency medicine databases. Key variables included trial design features, primary endpoint type and achievement status, and justification for approval in cases of failed efficacy endpoints.

Between 2020 and 2023, 227 new drugs were approved by the European Medicines Agency for 234 indications. Of these, 69 (30.4%) were granted orphan designation, and 53 (23.3%) were approved via an expedited pathway, most commonly conditional approval (29 drugs, 12.8%). Cancer was the leading therapeutic area, accounting for 59 approvals (26.0%). Approvals were supported by 404 pivotal clinical trials. Of the 227 drugs, 177 (78.0%) were supported by at least one randomized controlled trial, of which 138 (78.0%) were blinded. Out of those, 139 (78.5%) were supported exclusively by superiority trials. Regarding endpoint classification, 124/227 drugs (54.6%) relied exclusively on surrogate endpoints and 69 (30.4%) on clinical endpoints. Overall, 22 approvals (9.7%) were supported by at least one pivotal trial in which at least one primary endpoint was not met; in five of these cases (22.7%), the failed trial was the sole pivotal trial. The most common rationale for approval despite null primary results was reliance on the totality of evidence, secondary endpoints, or clinical judgment (nine products; 40.9%).

Our findings reveal substantial variability in the design and evidentiary strength of pivotal trials supporting European Medicines Agency approvals between 2020 and 2023. While the majority of studies were randomized controlled trials, reliance on surrogate endpoints was common. That 10% of approvals were based on pivotal trials with null primary endpoints highlights the nuanced role of regulatory judgment in therapeutic evaluation. These findings prompt reflection on evolving evidence standards in drug regulation and underscore the need for transparency and consistent justifications.

Bayesian methods allow trial investigators to combine evidence obtained within a clinical trial with relevant evidence that is available outside the trial. Bayesian analyses are now widely used in the drug development process, to inform internal ‘go/no-go’ decisions about planned studies, for example when deciding whether a drug should proceed from phase II to a phase III trial. However, Bayesian analyses are not commonly used for analysis of phase III (confirmatory) trials.

In this article, we performed a narrative review of confirmatory trials using Bayesian methods for their primary analysis, to explore which types of trials chose Bayesian methods, why they chose a Bayesian analysis and how the methods were used. We reviewed published papers over a 6-year period and explored the characteristics of trials using Bayesian methods for their primary analysis, their reasons for choosing a Bayesian analysis, whether any informative priors were used and if so how they were informed. Next, we selected four trials from the review as case studies and presented their motivation for using Bayesian methods and their Bayesian analyses in more detail.

Our narrative review found that the number of Bayesian methods in confirmatory clinical trials has approximately doubled over the past 6 years, reflecting growing familiarity among investigators. Ninety-four papers were eligible for inclusion, presenting results from 69 separate trials. The most common reason given for choosing Bayesian methods was to make direct probability statements about the superiority and/or futility of the interventions evaluated; this was mentioned for 49% of trials. Flexibility in adapting the design or use of Bayesian stopping rules was another very common motivation, cited for 47% of trials. Borrowing information through informative priors was cited for a much smaller proportion (16%) of trials. The majority of trials (75%) specified vague or weakly informative priors for all parameters.

Among the reasons given for choosing Bayesian methods, we consider the use of informative priors or making direct probability statements to be the strongest motivations for a Bayesian analysis, because there are no equivalent frequentist approaches. Making direct probability statements was the most common motivation provided, while informative priors were not often used. In settings with recruitment difficulties, we recommend considering borrowing relevant information, to gain power and precision. In all confirmatory trial settings, we recommend that Bayesian approaches are used only with careful justification, investigators make clear whether the methods and priors were pre-planned, and alternative frequentist approaches are considered.

Long COVID, the chronic sequela of SARS-CoV-2 infection, has affected millions of people worldwide. However, its pathogenesis and treatment remain unknown. To address this critical gap, the National Institutes of Health (NIH) developed the Researching COVID to Enhance Recovery (RECOVER) Initiative that included the Viral Persistence and Reactivation, and Immune Dysregulation (RECOVER-VITAL) study as one of the first of five NIH-sponsored, integrated platform protocols to support the rigorous and rapid investigation of potential interventions for Long COVID.

Experts across academia, the NIH, and the community of patients and caregivers were brought together to design the RECOVER-VITAL protocol. We present the challenges and rationale underpinning critical components of this protocol and provide evidence for use of the platform for efficient execution of current and future Long COVID clinical trials. Facets of the RECOVER-VITAL protocol, including inclusion criteria, intervention groups, study procedures, investigation of biomarkers, and endpoints, were carefully crafted to advance the current and future science and operations of Long COVID clinical trials, particularly in the absence of existing data. The initial trial within the RECOVER-VITAL platform evaluated two durations of an antiviral drug, nirmatrelvir/ritonavir, to test the proposed etiology of viral persistence on the Long COVID symptom clusters of autonomic dysfunction, cognitive dysfunction, and exercise intolerance. The primary outcome measures were patient-reported, and secondary outcomes included performance-based measures for each of the symptom clusters of interest.

The RECOVER-VITAL platform protocol has substantial implications for the design and conduct of future Long COVID clinical trials.

Although the analysis of event-based clinical trials commonly relies on assumptions about the underlying hazard functions, in practice it is rare to see estimates of those functions.

I describe conventional and novel methods for estimating the hazard function using discrete and discretized continuous survival models. The conventional approach involves parametric modeling; the novel approach applies Bayesian model averaging to flexible modeling by splines or fractional polynomials. I evaluate the methods in a Monte Carlo study and illustrate them in the analysis of three historical clinical trials.

Although flexible models can capture features of the hazard functions—such as multimodality—that parametric models miss, they are not foolproof. Spline modeling was generally the most reliable, in the sense of yielding good coverage probabilities for the mean and median with modest loss of efficiency. In the examples, the discreteness of the measurements—days, weeks, or months—had little effect on the shape of estimated hazard functions. All three data sets showed some evidence of departure from the proportional hazards assumption, but in only one did a test for proportionality detect this departure.

Flexible parametric models, estimated in the Bayesian model averaging framework, offer a robust approach to recovering the shape of the hazard function. Analyses of three clinical trial databases suggest that visualization of the hazard function can be a valuable adjunct to conventional survival analysis.

Informed consent forms are an essential component of ethical clinical research. However, they are failing. They are long, legalistic, and written at an 11th-grade reading level. Consequently, many participants have poor comprehension of the research they are considering. Our objective is to assess participant comprehension and satisfaction across three informed consent forms of different lengths and formats including one based on the Food and Drug Administration’s (FDA) 2024 draft summary template.

In total, 334 healthy participants from AmeriSpeak’s online survey panel were recruited via email and AmeriSpeak’s member portal, oversampling less educated individuals and those from racial and ethnic minority groups. Participants were randomized to three informed consent forms—long, medium, and short—and were asked to imagine hypothetically being a patient considering joining a study. They were given a brief introduction on the purpose of informed consent forms and of this project’s goal to evaluate comprehension, satisfaction, and willingness to join the described hypothetical study but were not told the hypothesis or that there were three different informed consent forms. The long informed consent form (21 pages) was used in an actual international Bristol Myers Squibb study concerning a stroke prevention regimen. The long informed consent form contained longer paragraphs and lists and was written just below an 11th-grade level. The long informed consent form was modified to create the medium (14 pages) and short informed consent forms (3 pages), which both used simpler words, shorter sentences, less passive voice, tables, and other patient-friendly edits to achieve below sixth-grade reading levels. The short informed consent form was adapted from the Food and Drug Administration’s 2024 draft summary template.

The medium informed consent form (mean score of 79.3%) scored significantly higher on comprehension than the long informed consent form (65.7% mean score, p < 0.001) or the short informed consent form (69.3% mean score, p = 0.006). Compared to the long informed consent form, fewer medium informed consent form participants found their informed consent form “too long” or “too detailed.” Compared to the long and medium informed consent forms, more short informed consent form participants reported wanting more information on study procedures.

The medium-length informed consent form, written at a sixth-grade reading level, with shorter words and sentences, tables, and enhanced formatting, produced significantly better comprehension than longer or shorter informed consent forms. The short informed consent form produced the same level of comprehension as a long informed consent form used in approved clinical trials. Researchers, pharmaceutical companies, and institutional review boards should adopt these informed consent form modifications to increase participant comprehension and satisfaction.

Non-inferiority trials can be used for efficacy endpoints or safety endpoints, with indirect and direct comparison to placebo. A less common application of this design involves using placebo as the proposed intervention to challenge standard practices that lack evidence of efficacy. However, the methodology of using a placebo in a non-inferiority trial is poorly described in the literature. We performed a scoping review to map how placebo is utilized in randomized controlled trials with a non-inferiority design, with particular attention to studies positioning placebo as a proposed alternative to existing, but unproven, interventions.

We conducted a scoping review of randomized controlled trials using non-inferiority designs with a placebo arm, searching six databases without date or language restrictions. Eligible studies were primary randomized controlled trials with at least one placebo arm evaluated under a non-inferiority hypothesis. Data extraction focused on study characteristics, design elements, rationale for non-inferiority design and margin, and analytical practices.

Of 6897 studies screened, 94 met inclusion criteria. Three primary study types were identified: safety (63%), deprescription (20%), and shorter-course (13%) trials. There has been increased use of deprescription and shorter-course studies since 2017. One-third (35%) hypothesized that placebo was non-inferior to active treatment, predominantly in deprescription and shorter-course trials focused on antibiotic use. Most studies (94%) applied non-inferiority analysis to primary outcomes, yet only 22% provided a rationale for non-inferiority design, and despite 96% prespecifying non-inferiority margin, only 41% justified the margin. While 71% used intention-to-treat analysis, only 53% conducted per-protocol analysis. Graphical representation of non-inferiority margins and confidence intervals was reported in 23% of studies.

Placebo is increasingly used in non-inferiority trials aimed at evaluating the necessity of standard interventions, including safety, deprescription, and shorter-course designs. However, many trials lack critical methodological transparency. Future studies should clearly justify non-inferiority designs and margins, use both intention-to-treat and per-protocol analyses, and adhere to the Consolidated Standards of Reporting Trials reporting guidelines to enhance interpretability and rigor.

Information regarding the use of wearable devices in clinical research, including disease areas, intervention techniques, trends in device types, and sample size targets, remains elusive. Therefore, we conducted a comprehensive review of clinical research trends related to wristband wearable devices in research planning and examined their applications in clinical investigations.

As this study identified trends in the adoption of wearable devices during the planning phase of clinical research, including specific disease areas and targeted number of intervention cases, we searched ClinicalTrials.gov—a prominent platform for registering and disseminating clinical research. Since wrist-worn devices represent a large share of the market, we focused on wrist-worn devices and selected the most representative models among them. The main analysis focused on major wearable devices to facilitate data analysis and interpretation, but other wearables were also surveyed for reference. We searched ClinicalTrials.gov with the keywords “ActiGraph,”“Apple Watch,”“Empatica,”“Fitbit,”“Garmin,” and “wearable devices” to obtain studies published up to 21 August 2022. This initial search yielded 3214 studies. After excluding duplicate National Clinical Trial studies (the overlap was permissible among different device types except for wearable devices), our analysis focused on 2930 studies, including simple, time-series, and type-specific assessments of various variables.

Overall, an increasing number of clinical studies have incorporated wearable devices since 2012. While ActiGraph and Fitbit initially dominated this landscape, the use of other devices has steadily increased, constituting approximately 10% of the total after 2015. Observational studies outnumbered intervention studies, with behavioral and device-based interventions being particularly prevalent. Regarding disease types, cancer and cardiovascular diseases accounted for approximately 20% of the total. Notably, 114 studies adopted multiple devices simultaneously within the context of their clinical investigations.

Our findings revealed that the utilization of wearable devices for data collection and behavioral interventions in various disease areas has been increasing over time since 2012. The increase in the number of studies over the past 3 years has been particularly significant, suggesting that this trend will continue to accelerate in the future. Devices and their evaluation methods that have undergone thorough validation, confirmed their accuracy, and adhered to established legal regulations will likely assume a pivotal role in evaluations, allowing for remote clinical trials. Moreover, behavioral intervention therapy utilizing apps is becoming more extensive, and we expect to see more examples that will lead to their approval as programmed medical devices in the future.