Novel Technologies for Identifying Treatment Assignment in a Prehospital Pediatric Emergency Care Clinical Trial

Abstract

Clinical trials are essential for generating evidence-based knowledge to support clinical practice. In the prehospital setting, particular challenges exist in clinical trial design that have a large impact on the quality of the results, including the appropriate and systematic allocation of assigned treatment arm. The Pediatric Prehospital Airway Resuscitation Trial is a multicenter clinical trial evaluating different paramedic airway management strategies on critically ill and injured children in the prehospital setting. The trial assigns airway management strategy using odd-even day pseudo-randomization, which pragmatically balances the rigor of a statistical design that minimizes bias with the complexities of operationalizing a clinical trial in the prehospital setting. We describe the development of 2 technologies to aid in treatment assignment adherence during the trial: (1) a quick response code-linked web application, and (2) a self-contained portable reminder device. These technologies offer novel approaches to support treatment allocation in acute care clinical trials.

1. Introduction

Clinical trials are essential for generating knowledge to guide clinical practices.¹ Random allocation of study interventions is a hallmark of clinical trials, balancing confounders, minimizing bias, and enhancing the validity of the results.² Maximum adherence with treatment assignment is essential in a clinical trial to ensure that the study results in valid inferences.

Investigators employ a wide range of methods to facilitate random allocation. For example, studies of investigational drugs may use blinded vials containing either study drug or placebo.^3,4 Research teams may learn of treatment assignment through sealed blinded envelopes.⁵ In contemporary practice, research personnel commonly access randomization assignments through an online website or the study database.⁶ Random allocation is especially challenging in trials involving life-threatening conditions that enroll participants through exception from informed consent. Examples include cardiac arrest, severe hemorrhage, major trauma, respiratory failure, and stroke. Delivering patient care in these settings must be initiated without delay, often within seconds to minutes, not allowing for the use of traditional allocation methods.

One strategy used for treatment allocation in clinical trials is time-based allocation, a form of pseudo-randomization where study treatments alternate between defined time periods.⁷ For example, in the Resuscitation Outcomes Consortium Continuous Chest Compressions trial, Emergency Medical Services (EMS) agencies alternated between a compression-to-ventilation ratio of 30:2 and continuous chest compressions in 6-month cycles.⁸ In the Los Angeles County pediatric airway trial, paramedics alternated between strategies of bag-valve-mask (BVM) ventilation only and BVM followed by endotracheal intubation on an odd-even calendar day basis.⁹ Though allocation is not concealed with time-based allocation, it is appealing for life-threatening conditions in the prehospital setting because it is relatively simple, can be quickly executed, and is preassigned before the patient encounter.

An important challenge of time-based allocation is that care teams may not remember the treatment assignment for the specified time period. The Pediatric Prehospital Airway Resuscitation Trial (Pedi-PART) is a national multicenter trial comparing paramedic airway management strategies in critically ill and injured children in the prehospital setting.¹⁰ Pedi-PART uses time-based allocation, assigning patients to study treatments on an odd-even calendar day basis. We describe the development of 2 digital technologies to aid paramedics in remembering the assigned intervention each day in the Pedi-PART trial.

2. The Trial and Treatment Allocation Method

Pedi-PART (NCT 06364280) is a multicenter clinical trial comparing strategies of paramedic prehospital airway management.¹⁰ The trial involves EMS agencies from 10 United States (US) cities and surrounding metropolitan areas: Austin, TX; Charlotte, NC; Colorado (Denver, Aurora and Colorado Springs, CO); Columbus, OH; Indianapolis, IN; Los Angeles, CA; Milwaukee, WI; Sacramento, CA; Tucson, AZ; and Washington, DC. The trial includes children ages 1 day through 17 years old requiring prehospital airway management due to cardiac arrest, traumatic injury, or other causes of respiratory failure. The trial interventions include 3 primary airway management strategies: (1) bag-valve-mask ventilation alone (BVM), (2) supraglottic airway (SGA) placement, and (3) endotracheal intubation (ETI). The primary outcome is 30-day intensive care unit (ICU)-free survival. Pedi-PART’s adaptive design prespecifies periodic interim analyses for early success and includes 2 distinct stages: Stage I compares BVM-alone with a primary SGA placement strategy, and stage II compares the winning strategy from stage I with a primary ETI strategy. Up to 3000 participants will be included in the trial.

Pedi-PART assigns treatments through odd-even calendar day time-based allocation. On odd numbered days in stage I of the trial, EMS personnel use a primary SGA placement strategy, where BVM is followed by SGA insertion. On even numbered days in stage I, EMS personnel use a BVM-only strategy when only BVM is used. On the 31^st and 1^st of the month, EMS personnel adhere to odd day assignment. Because many EMS agencies work 24-hour shifts, the investigators asked each EMS agency to preselect an anchor time for switching treatment assignments each day for the duration of the trial. For example, an agency may choose to switch between odd-even day assignments based on an anchor time of 7:00 AM. Thus, for an incident when the EMS crew arrives on-scene at either May 1, 8:00 AM or May 2, 4:00 AM, EMS personnel would be allocated to the odd day assignment (primary SGA strategy) because both fall within the 24-hour period of when the odd day assignment began for that agency. But for incidents when EMS crews arrive on-scene at May 2, 7:00 AM, or up to 24 hours after that, they would use the even day assignment (BVM-only strategy). The investigators instructed EMS agencies to use on-scene arrival, rather than dispatch time, as the basis for the anchor time for treatment assignment since that is when the EMS crew’s clinical treatment begins. In some cities where units from multiple EMS agencies can be present on-scene, the EMS personnel were instructed to use the treatment allocation based on the anchor time for the first unit to arrive on-scene.

Multiple factors motivated the selection of time-based allocation by odd-even calendar day. EMS personnel and research teams discouraged the use of sequentially numbered sealed envelope randomization as envelopes are prone to get lost, would be difficult to restock, and could result in greater anxiety for the EMS clinicians who must rapidly care for a critical child. Investigators ruled out contacting a call center or accessing a web page as potential allocation strategies given the unreliability of cellular or wireless signals in the prehospital setting and the time-critical nature of cardiac arrest, major trauma and respiratory failure. Similar to the Pragmatic Airway Resuscitation Trial (PART) and other trials of the Resuscitation Outcomes Consortium, the investigators considered using a cluster randomization approach using cross-over time periods of 3-6 months for each intervention.^8,11,12 However, because the Bayesian adaptive design of Pedi-PART involves frequent interim analyses (every 300 patients enrolled), prolonged cross-over periods would likely result in considerable imbalance between treatment groups. In addition, an odd-even day approach more closely approximates a randomized control trial in its ability to minimize bias.

Odd-even day treatment assignment was successfully employed in Gausche-Hill’s prior clinical trial of pediatric prehospital airway management, resulting in an intervention compliance rate of 90%.⁹ The odd-even day approach aligned well with EMS shift structures in the consortium (most agencies work 24-hour shifts) and could accommodate variations across a given EMS agency; for example, application of varying anchor times across EMS agencies. This approach also mitigated concerns with seasonal disease variations that might bias monthly or quarterly alternation. The odd-even day alignment with EMS shift structures combined with the emergent nature of the conditions treated in Pedi-PART; however, reduce the likelihood of this bias. A priori input from EMS personnel from participating sites affirmed that odd-even time-based allocation was the best and most feasible allocation strategy for Pedi-PART.

3. Technical and Operational Requirements

Given the unique challenges of conducting prospective research in the prehospital environment, the stress of caring for critically ill children, and the limited frequency of pediatric airway events, EMS personnel might have trouble remembering the odd-even day treatment assignment, even with ample training. In addition to multiple visual prompts placed on EMS units (eg, posters and cue cards), novel portable technologies were considered as potentially more effective reminders of the treatment assignment. Several key operational considerations for such novel technologies include:

• -Usability—Any device must be simple to use, minimize disruptions in clinical care, and account for potential limitations of dexterity in the prehospital environment.
• -Platforms and Connectivity – EMS personnel often have access to smart phones and laptop computers. Program “apps” are familiar and can be easily deployed on portable technology. However, some EMS agencies do not allow use of personal smartphones, and some personnel will not use personal devices on duty. Furthermore, device connectivity may be limited, especially in rural settings. Therefore, the technology must be small, portable, and able to function independently of internet or cellular signals.
• -Functionality—The technology must display the treatment assignment in real-time for any given date and time. It must be programmable for different EMS agencies and must accommodate daylight savings time changes and leap year date changes. The technology must accurately track dates and times without an internet connection. It must exhibit minimum time drift over the projected 4-year project period.
• -Portability and Durability—Any device must be portable and durable and able to fit in an EMS airway kit. It must maintain reliability in varying environments, including extremes of high (Arizona) and low (Milwaukee) temperatures as well as varying levels of humidity. The device must be waterproof and rugged to endure handling in the prehospital environment. It must be easily viewed in dark and bright conditions. Collecting devices to perform reprogramming or recharging is logistically difficult, so the technology must have a long battery life, minimizing the need to recharge to no more than once every 6 months. The device must be rechargeable with existing standard adapters and connections, such as USB-C.
• -Programmability—The technology must be easily updated or reprogrammed. This allows the device to be updated for Stage II of the trial, as well as to be used in future prehospital clinical trials.
• -Economical—The cost of development and production must meet budgetary limits (ideally under $30 per EMS unit). We estimated needing 2500 units to accommodate all Advanced Life Support units in the study. Production and deployment of this technology must meet the rigorous onboarding schedule for the trial.

We considered several options to satisfy the above requirements, including but not limited to a custom network-independent device, native phone application, and a web application viewable on any modern device, including phones, tablets, and laptops commonly used by EMS. The combination of a network-independent device with a web application covered all major requirements described. The native phone application was determined to be the least desirable of the options as it would likely require the use of personal devices which are restricted by some agencies and the potential unwillingness of other EMS personnel to use personal devices.^13,14 Although phone applications would satisfy maintaining time without continuous network access, they introduce their own challenges. Native application development is costly compared with web application equivalents due to the multiple platform variations (ie, various forms of iOS and Android), requiring development and maintenance of multiple code bases, as well as deployment of updates through a third-party app store, which delays deployment.^15,16

Based on these considerations, we developed 2 technologies for the Pedi-PART trial.

4. Novel Technologies

4.1. Research Assignment Learner for the Prehospital Setting (RALPH)

RALPH is a portable standalone device inspired by computer “key fob” technology (Figs 1 and 2). We developed the prototype for RALPH using a SAMD21 (Seed Studio Xiao SAMD21) development microcontroller and an independent DS3231 real-time clock (RTC). A commercial partner (Anzer, Akron, OH) modified the prototype for large-scale production, using a programmable STM32G07 processor (ST Microelectronics) with an onboard RTC. When activated, the device displays the current date and time, and the treatment allocation for the EMS agency. The individual units are programmed to account for varying EMS agency time zones and study anchor times. The device can be recharged or reprogrammed using a standard USB-C port. The device does not require cellular or wireless signal access.

Figure 1.

Research Assignment Learner for the Prehospital Setting.

Figure 2.

Software cycle for Research Assignment Learner for the Prehospital Setting device. OLED, organic light-emitting diode.

RALPH tracks time using a 32.768 Khz quartz oscillator clock. A binary counter in the unit’s STM32G07 processor divides the clock output to indicate once per second intervals. When initially programmed, the device stores the clock time from the US National Institute of Standards and Technology (NIST). The RALPH device calculates Julian date and local time based upon the number of seconds elapsed since initial programming of the device, accounting for daylight savings time when applicable based on the deployment region. RALPH stores the programming time, programming date, shift link, study phase data offset, display on time, shift start time, and shift length in a battery-backed random access memory (RAM). When the user presses the button on the RALPH device, the software recalls the data from RAM, calculates the display date and time, and runs a comparison calculation against the stored data to determine which study day (odd or even day) to display.

Battery longevity was a priority in device development. By storing only the elapsed number of seconds and only calculating data on button activation, the device’s “standby” current consumption is low, encompassing 18 microamperes (mA)for the processor and 4 microamperes for the quartz oscillator crystal. When activated, the display screen consumes 10 mA of current at 3.3 V, reverting to standby after 20 seconds of display. Using a 4.2 V 1500 mAH lithium polymer battery, the device has a minimum life of 6 months between charges.

The housing of the RALPH device is constructed from Acrylonitrile Butadiene Styrene plastic and is designed to sustain vibrations and moisture exposure in the prehospital setting. The device is compact (78 × 48 × 25 mm) and lightweight (70 g), allowing a range of options for attaching to bags or placement in EMS airway kits. The device has a 25 x 5 mm back illuminated white on blue colored organic light emitting diode display. We conducted electromagnetic field interference testing on early production devices to ensure that the device would not interfere with any medical devices in the ambulance environment. The components and enclosures were tested to withstand environmental requirements.

4.2. QR Code-Based Emergency Research Reminder Technology (QBERRT)

QR codes are commonly used to automate loading of website uniform resource locators (URL) on smart phone devices. To provide additional options for study EMS agencies, we created QBERRT (Fig 3). When the EMS personnel scan the QR code with a standard mobile device camera, the user device displays a website with the research intervention assignment for a given EMS agency, date and time. We developed QBERRT on a cloud-based server to improve accessibility across the US. We used the same logic as in the RALPH device to determine and display the appropriate intervention for any combination of date and time, except that this system uses network time protocol to obtain accurate times from standard NIST internet time servers rather than relying on an onboard RTC. We developed separate QR codes for each of the 10 enrolling sites, accounting for the different time zones and anchor times. Once accessed, the device displays the current time, expected time zone based on QR code used, and the appropriate assignment to match with associated graphical image of the device to use.

Figure 3.

QR Code Based Emergency Research Reminder. Pedi-PART, Pediatric Prehospital Airway Resuscitation Trial.

EMS agencies have deployed QBERRT using a range of strategies. Some have placed their site QR codes on laminated stickers in airway kits. Some have saved the URL as a bookmark on laptop or smartphone devices. Some have used the booked programmatically generated images onto existing EMS dashboards or kiosks in EMS stations.

5. Discussion

Keeping care teams informed of research trial assignments can be difficult with pseudo-randomization. We describe the development of 2 novel technologies to support execution of the Pedi-PART trial. These technologies have strong potential to positively influence trial treatment adherence. If successful, this would support the use of self-contained portable technologies in the design and implementation of similar clinical trials in acute care settings.

Two technologies were developed to address the complex requirements described with clear compromises between them. In comparing these solutions, RALPH provides a technology independent of internet connectivity and the potential need for EMS personnel to use personal devices compared with QBERRT. The latter, however, is easier to reprogram and deploy where these requirements can be met while also being less expensive and less prone to time drift given the ability to regularly synchronize time with standardized NIST internet time servers, effectively eliminating time drift common to embedded hardware solutions.

Our early experience with RALPH and QBERRT has yielded key practical lessons. Significant clock time drift occurred with early production versions of RALPH. We solved the issue by replacing the oscillating crystal with an ECS International model ECS-327 temperature compensated quartz oscillator module to greatly improve timekeeping accuracy and thus study validity. Deployment of portable devices can be challenging as some EMS agencies have dozens—if not hundreds—of vehicles. Battery life has been longer than expected, with some devices showing negligible battery drain after 6 months. There was universal agreement to deploy such strategies for the Pedi-PART trial by EMS site investigators; however, we found that not all EMS agencies embraced the portable device; several declined to deploy the devices, reasoning that they were more accustomed to using rote memory to recall the odd-even day assignment.

5.1. Future Directions

Our demonstration sets the stage for future technology to support the implementation of acute care clinical trials. Per-patient randomization is desirable in clinical trials as it results best in balanced allocation between treatment groups. We aspire to evolve the RALPH device to incorporate cellular or wireless technology, enabling real-time connection with a clinical trial data coordinating center and patient-level randomization. Several challenges need to be overcome to meet this goal. Portability is clearly important in the prehospital setting. Any devices using cellular or wireless technology must meet the same size and weight standards of the current RALPH devices. These devices must be able to function well with marginal or weak cellular or wireless signals. They must also have sufficient battery life. Given that a single EMS agency may have dozens (and perhaps hundreds) of response vehicles, retrieving devices for recharging requires significant effort.

These devices must not interfere with EMS medical devices such as portable cardiac monitors, defibrillators or portable radios. Finally, cost is always a consideration. Cellular and wireless-enabled devices may not be feasible if their cost exceeds available research development resources. Lastly, we have observed some use of the QBERRT technology, which suggests that with wider and more reliable cellular and wireless availability, it may be possible to rely on existing technologies carried by EMS personnel through development of a native phone application maintained along with a web application for higher accessibility.

An important consideration of any technology supporting a clinical trial is whether it effectively accomplish its goals. The underlying question for this trial is whether the novel technologies result in improved adherence with odd-even day treatment assignment. As the trial is ongoing, this question cannot be currently assessed but may be able to be appraised after the conclusion of the study.

6. Conclusion

In conclusion, we describe the development of novel technologies to aid with allocation adherence in prehospital acute care trials and identify challenges for future development goals. Development of these technologies may help to enhance treatment allocation adherence in prehospital acute care trials.

Author Contributions

Conceptualization: HEW, TPS, JAF, KM, JM.

Funding acquisition: HEW, JVB, MGH, MIS, MH, NB.

Methodology: TPS, HEW, KM.

Project administration: HEW.

Resources: HEW, TPS, JAF, JM.

Software: TPS, JM.

Supervision: JF, KM.

Writing – original draft: TPS, HEW.

Writing – review and editing: TPS, JM, KM, JAF, JMV, MGH, MIS, NB, MH, BW, HEW.

Funding and Support

This study was supported by the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health (NIH) under awards UG3HL165019 and U24HL165014. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NHLBI/NIH. This study was conducted in the Pediatric Emergency Care Applied Research Network (PECARN). PECARN is supported by the Health Resources and Services Administration (HRSA) of the US Department of Health and Human Services (HHS), in the Maternal Child Health Bureau (MCHB), under the Emergency Medical Services for Children (EMSC) program through the following cooperative agreements that were in place when the data was collected and analyzed: CHaMP node – The Leland Stanford Junior University (U03MC54727); EMSC Data Center-University of Utah (UJ5MC30824); GLACIER-Nationwide Children’s Hospital (U03MC28844); HOMERUN-Cincinnati Children’s Hospital Medical Center (U03MC22684); PEMNEWS-Columbia University Medical Center (U03MC00007); PRIME-University of California at Davis Medical Center (U03MC00001); SPARC- Rhode Island Hospital/Hasbro Children’s Hospital (U03MC33155); STELAR – Seattle Children’s Hospital (U03MC33156). This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by HRSA,HHS or the US Government.

The funding sources had no role in the study design, data collection, analysis and interpretation of data, writing of the report, or decision to submit the article for publication. This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by HRSA, HHS, or the US Government.

Conflict Of Interest

Dr. Wang receives grant support from the National Institutes of Health and Department of Defense, research funds as a site PI for a study by Vasomune, Inc. Dr. Hansen receives grant support from the National Institutes of Health and the Health Resources and Services Administration. Dr. Shah receives grant support from the National Institutes of Health and the Health Resources and Services Administration. The rest of the authors have nothing to report.