An Innovative Approach to Building an Effective and Efficient Pediatric Trial Network

Conducting randomized controlled trials (RCTs) in children is exceptionally challenging because of low disease burden, ethical considerations, and limited funding.¹  The number of RCTs in children are few, representing an urgent and unmet need. Coon et al describe an approach to RCTs in hospitalized children characterized by dispersed low-volume recruitment, incentives, pragmatic questions, streamlined regulatory process, and simultaneous conduct of multiple trials.²  They conclude that the High Efficiency RandOmIzed Controlled (HEROIC) approach will result in fast, agile, cost-effective, and generalizable RCTs.

The HEROIC approach is promising in several ways. Dispersed low-volume recruitment is successful in other networks including the Pediatric Trials Network (PTN).³  The PTN comprises >200 sites across North America and provides scientific, technical, and administrative infrastructure necessary to conduct safe, effective, and efficient trials in children. Over the last 12 years, the PTN has used a dispersed recruitment model to enroll >11 000 children across 44 studies and complete 26 data submissions to the United States Food and Drug Administration leading to 17 Food and Drug Administration label changes.³ 

Coon and colleagues envision recruiting participants from across a large network of geographically dispersed sites and nontraditional clinical trial sites, such as children hospitalized at a general community hospital, to improve inclusion and diversity.⁴  We agree, based on our experience, that this may help address longstanding equity issues and underrepresentation of diverse populations in trials.⁵ 

To build the infrastructure necessary for a large, diverse, and efficient network, Coon and colleagues describe a single data use agreement with core data elements to cover multiple trials and master reliance agreements to execute multiple trials per year. Similarly, we use master alliance agreements in the National Institutes of Health–funded Trial Innovation Network to establish reliance and support the regulatory process across the nation.⁶  Although these tactics are helpful, we strongly suggest the use of master protocols.

Master protocols allow multiple interventions and/or populations to be evaluated simultaneously under a single protocol and have emerged as a strategy to reduce time and costs associated with developing and delivering new interventions. We use a master protocol for the PTN Pharmacokinetics of Understudied Drugs Administered to Children per Standard of Care study (NCT04278404), which includes participants from rural and medically underserved communities among a target population of 5000 infants and children evaluating >70 commonly-used therapeutics. Our group was the coordinating center for 2 Trial Innovation Network placebo-controlled trials that successfully used master protocols to evaluate several therapeutic agents for severe acute respiratory syndrome coronavirus 2 infection (ACTIV-1, NCT04593940 and ACTIV-6, NCT04885530).⁷ 

Another clear strength of the HEROIC approach is the focus on pragmatic questions, which we anticipate will be key to the network’s success. Designing pragmatic trials with research questions selected by the investigators is especially important for financial efficiency but can be challenging. To prioritize the important questions proposed by the authors, we offer advice based on our experience:

1. Evaluate the degree of pragmatism of each question and scale efforts appropriately. Many trials labeled as “pragmatic” are not at all pragmatic,⁸  which increases sample size and reduces relevance. Tools such as PRECIS-2 (https://www.precis-2.org) can assess the extent to which RCTs are pragmatic and relevant.

2. Begin by studying interventions in the “pre-competitive space” and, therefore, not subject to patent or intellectual property conflict. Examples include off-patent drugs and devices (in which we have done much of our work), as well as oxygen delivery, ventilator settings, and intravenous fluids. Although on-patent products should certainly be studied, funding pathways and regulatory guidance add both complexity and expense.

3. Ensure that the research question fits the sites. Conducting trials at sites with low burden of disease, or low event of the primary outcome, is inefficient and should generally be avoided in the early stages of the network. As the network grows, it can target sites with the desired prevalence of disease or event being studied.⁹  We have previously partnered with other National Institutes of Health–sponsored research groups and private-sector groups to complete pretrial epidemiology.¹⁰ 

4. Finally, begin with a small, pragmatic trial with limited data collection, through which processes, procedures, and workflows can be developed and honed before scaling to larger, more complex trials. Researchers learn to how to conduct trials together by conducting trials together. The first PTN trial (NCT01222585) enrolled 24 infants at 3 sites.¹¹  We have since enrolled >11 000 children and studied 132 molecules. In the early stages of the HEROIC network, we recommend focusing efforts on completing the first trial. The success of the first trial should be measured not by the P value, but by the fraction of investigators who return for the second trial.

Incentives are important tools for engaging investigators and building a successful network. The authors wisely propose using a mix of financial support, authorship, research training, and/or credits for maintenance of certification to incentivize site participation, but they will need to consider specific details. Regarding financial support, some dispersed research networks provide a relatively small base amount of financial support with higher weight toward per-patient enrollment. This strategy is cost-effective but could be difficult for sites without existing research infrastructure.^12,13  Authorship can be a powerful incentive but may be difficult to extend to all investigators in a dispersed model in excess of 50 sites. Group and consortium authorship can allow additional coauthors to be indexed in searchable interfaces such as PubMed. The HEROIC network proposes a flexible, open structure, which offers important advantages in terms of diversity, but may find practical limitations to the time, resources, and effort that individual sites and investigators can devote to its work. One suggestion is to draw in features of exclusive networks, such as internal opportunities for funding and leadership positions, to incentivize participation. We also found maintenance of certification credits bolstered participation across 19 sites in a trial of sedation regimens.¹⁴  Because different incentives motivate different investigators, enrollment into the HEROIC network will likely benefit from a combination incentives as described. We also recommend engaging site investigators with individualized conversations, in person or by phone, to understand the optimal mix of incentives for the HEROIC network.

One potential challenge we anticipate for the HEROIC network is administrative burden. Dispersed recruitment over 50 to 150 sites requires substantial administrative effort to identify, initiate, and coordinate investigators and staff. Inexperienced sites that lack preexisting research infrastructure often face limitations in staff capacity and training. Although single institutional review boards (IRBs) are designed to streamline the regulatory process and are necessary for very large, complex, international trials like ACTIV-1, critics argue that single IRBs simply shift the administrative burden from local IRBs to already-burdened investigators.¹⁵  We have seen evidence to support this criticism at some centers, which we believe is because single IRBs still require ‘local context’ reviews by sites, a process that can be lengthy and delay trial start-up.¹⁶  Preparing for single IRB review can also add costs to sites and the overall study.¹⁶  Inexperienced sites again are vulnerable to burdensome requirements. HEROIC trials risk falling into the divide of “have” and “have nots” of clinical research, in which success is determined by preexisting research infrastructure, administrative and support staff, and other resources. We recommend any new network study the process during their first several trials, from site identification and initiation through enrollment and dissemination, to learn how to support sites, maintain diversity, grow the network, and plan future large-scale research efforts.

RCTs may be viewed by some stakeholders to be less relevant in an era of novel study designs. Although alternative approaches to RCTs (such as studies embedded in electronic health records and administrative databases,¹⁷  direct-to-family trials,¹⁸  and opportunistic trials using scavenged samples¹⁹ ) can reduce burden and costs, RCTs will remain central to public health.¹⁷  Clinicians and regulatory agencies often want or require RCTs for clinical decision-making and approval of new therapeutics, respectively.^20,21  However, when possible, RCTs should incorporate features of novel study designs that have proven to be successful. The network should remain open to modifications on the basis of lessons learned in early stages and trial innovations that emerge in coming years.

In sum, the HEROIC approach offers great promise for conducting effective and efficient RCTs in hospitalized children. Based on our experience leading several large trial networks, we recommend organizing the HEROIC features around these specific strategies: (1) begin with a small pilot trial to build successful workflows and establish unity; (2) use master protocols to expand participation and increase efficiency; (3) focus on truly pragmatic questions and simple design to facilitate enrollment; (4) conduct pretrial epidemiology to select appropriate sites; (5) support inclusion of diverse and nontraditional sites; (6) use a mix of incentives; and (7) streamline regulatory and administrative processes as much as possible, while being mindful of how new processes can create additional burdens.

Those of us who provide clinical care to children have an ethical obligation to provide optimal care, which requires the execution of RCTs. Every day we delay RCTs is another day we subject children to off-label therapeutics, understudied interventions, and iatrogenic harm. HEROIC investigators have developed a thoughtful and promising approach to conducting RCTs. We look forward to their future work evaluating interventions and improving care for hospitalized children.