Respiratory Research in the Critically Ill Pediatric Patient: Why Is It So Difficult?

Study Design

The research design must be robust to accommodate the clinical environment and answer the research question. Core design elements, embedded in the research question, include delineating the patient situation/problem and a clinically important outcome.

Inclusion and exclusion criteria are crafted to describe the desired patient situation of concern. The “pediatric” age group expected to be enrolled in a clinical trial may appear obvious, but trialists have struggled with the operational definition of the “pediatric” patient. Pediatric ICUs typically admit patients across the age and developmental spectrum. Pediatric trialists enroll infants, not expected to exhibit neonatal physiology, through adolescents, not expected to exhibit mature physiology. Anticipating a developmental impact on a study's outcome, all age groups are typically included in pediatric RCTs, then age cohorts are controlled statistically during the data analysis phase. This improves the generalizability of study results to the entire PICU population.

The etiology of ALI/ARDS is diverse, and although substantial progress has been made in operationalizing ALI and ARDS, targeting interventions to specific disease entities may be beneficial. Patients with markedly different pathophysiology, for example, patients with airway obstruction, cyanotic heart disease, progressive neuromuscular weakness, and pulmonary hypoplasia should be studied as separate diagnostic subgroups.⁵ Patients with direct/indirect ALI can be enrolled into a pediatric RCTs, then disease cohorts can be evaluated in post-hoc analyses.

Using mortality as a primary outcome variable is not practical in pediatric ALI/ARDS trials.⁴ An impossible number of subjects would be required to demonstrate effect. Attributing death to ALI/ARDS, as opposed to underlying disease, is also difficult, as most children with ARDS have comorbid conditions that are unique and different than those of adults. Mortality is also insensitive to other important clinical outcomes, for example, survival with substantial neurologic morbidity.

Alternative end points such as ventilator-free days¹⁷ are used, and under appropriate circumstances, major morbidities could also be considered as primary end points.¹⁸ Measures of morbidity include composite end points such as time-to-resolution of organ failure. To support this work, validated organ dysfunction scores have been developed for children.^19,20 In addition, we know that adult survivors of critical illness suffer a decline in health-related quality of life²¹ but comparable pediatric data do not exist. Describing long-term outcomes, including functional²² or neurocognitive health, quality of life, and economic burden after pediatric critical illness, would also benefit the field.

Subject Enrollment

There is substantial variation in patient populations across clinical sites.⁵ Knowledge of local referral patterns is essential when recruiting site participation in clinical trials. Data points can include a PICU's recent history of subject enrollment in a clinical trial enrolling a similar patient population, or asking the site to complete mock screening for at least one month. Although seasonal variation will impact screening in any month, asking a site to prospectively apply inclusion/exclusion criteria allows a fair representation of the projected enrollment from candidate sites. These data, collected in a de-identified manner, are considered preparatory for research and, depending on the local institutional review board (IRB), may be considered an activity that does not require local IRB approval.

All sites participating in a multi-site clinical trial must use the same protocol. Permitting substantial variation in what happens to subjects at different sites introduces bias and risks, rendering study results uninterpretable. The system for protecting research subjects was designed when most research studies took place at a single institution. With multi-site clinical research, the same protocol is reviewed multiple times. Menikoff²³ noted that these duplicative reviews provide relatively few benefits and may reduce the likelihood that studies are in keeping with relevant ethical standards. To limit protocol drift between participating sites, coordinating centers should provide each participating sites with a template of a completed IRB application and informed consent form. Site co-investigators can then model their application and informed consent form on the template. All subsequent changes are reviewed and approved by the coordinating center.

Approximately half of all eligible patients are enrolled into clinical trials. The primary reason for this eligibility-enrollment gap is parent or legal guardian refusal.²⁴ Parents are ethically entitled and legally empowered to make decisions on behalf of their children.²⁵ There is site-to-site variability in consent rates, based on PICU and investigator-specific characteristics. Obtaining informed consent is difficult, especially when enrollment is time-sensitive. The parent or legal guardian may not be physically or emotionally available. Jansen and others have argued that deferred consent may be justified in emergency critical care research.²⁶

Federal regulations also require that the child, when cognitively capable, provide his or her assent to participate in research. While the age of assent varies on a local level, most IRBs require a plan on how assent will be obtained when enrolling children older than 7 years of age. When obtaining assent, the research and potential risks and benefits are explained using language that a child can understand. The readability statistics available on most word processing programs can help investigators level language to match the subject's age and cognitive capacity.

Protocol Adherence

Ensuring protocol adherence is critically important in multi-site clinical trials. Inter-site variation threatens a study's internal validity. All potential co-investigators, and their colleagues, must agree to follow the study protocol. Prior to inserting a protocol into a clinical environment, assumptions about the research environment require validation. Practice standards may vary across clinical settings, especially international settings. Practice surveys that include data elements that may impact a study's primary outcome should be completed at the start of, and periodically throughout the clinical trial. For example, if the primary outcome variable for a study is ventilator-free days, clinical trialists will want to know about each participating site's baseline ventilator-associated pneumonia (VAP) prevention practices and adherence rate, because VAP rate will impact ventilator-free days.

Subject safety requires that all clinicians demonstrate competence in the clinical protocol before enrolling subjects. Since different disciplines interact with a clinical protocol in varying ways, discipline-specific protocol “certification” and competence should be clearly established. Competence can be maintained by enrolling a minimum number of subjects each quarter, or by retraining.

Elements of a protocol that are critical to the validity of the study's conclusions are monitored and reported by site on a quarterly basis. Reports can be automatically extracted from existing case report forms, or specific queries can be sent to participating sites on a random basis. All primary outcome data require source verification, which is completed during site visits.

Sustained Commitment

Clinical research teams work to sustain the commitment and collaboration of multiple disciplines over the study period. Sustaining enthusiasm over a long clinical trial is challenging. Conference calls help to bring investigators together to review study progress and collectively solve problems and address competing demands.

It is important to choose one's collaborators wisely. Collaborators should be experienced in clinical research and should be local leaders who can champion the project. Local investigators should have the ability to devote time to the clinical trial and have a history of successful collaboration. Potential collaborators should provide evidence of their capacity to enroll and safely manage a critically ill patient on a protocol. Optimally, each site should have a skilled study coordinator who is knowledgeable in the field, detail-oriented, tenacious, and single-minded.

Undoubtedly, clinical trials will overlap.²⁷ Table 1 reviews factors that are considered when evaluating the capacity of several studies to enroll subjects from a single site. Intervention trials in the same patient population are avoided. When appropriate, sharing schemes are first agreed upon by each study's principal investigator. Recommendations for sharing schemes are then proposed to the clinical sites to ensure consistent decision making across all clinical sites. This process requires monitoring to ensure that the agreements are successfully implemented.

Including Children

Fundamentally, we should challenge the convention of automatically excluding children from clinical studies conducted in adult or neonatal patients. Pediatric patients should be excluded when there are developmental differences and included when there are not.⁵ In 1998, the NIH mandated that investigators include children in adult trials unless their exclusion was clearly justified.²⁸ Although several RCTs in adults have extended eligibility criteria to children 12 years of age, this maneuver has not provided the field with any new data to improve the care provided to pediatric patients with acute respiratory illnesses.^29,30

Halpern and colleagues have challenged the notion that novel interventions be always tested in adults prior to children.³¹ They note that, although the practice may protect children from research risks, it also may threaten a child's well-being by depriving them of evidence to guide their care. Children are traditionally viewed as “participants of last resort,” due to their vulnerability and decisional incapacity. Halpren et al note that RCTs should be conducted in children first when children are more apt to benefit and adults are more apt to be harmed.

Interest continues in identifying reliable biomarkers to both detect ALI and pace ALI resolution in response to a therapy.^32,33 Ideal biomarkers that are sensitive and specific, reflecting disease progression and/or resolution, have yet to be validated. Similar to targeted therapy in oncology, Angus and others note that we are close to using theragnostics in critical care: specifically, using biomarkers to identify patients most likely to respond to an intervention.³⁴ Once validated, we may be able to use biomarkers to adequately power smaller more feasible pediatric clinical trials.

Restructuring the Clinical Research Enterprise

The NIH recently highlighted the need to reinvigorate the biomedical research community through one of 5 key themes of the Director's Opportunity for Research award (RFA-OD-10–005). The award encouraged investigators to cultivate new collaborations by assembling multidisciplinary or interdisciplinary teams in conducting innovative research. The goal of the award was to strengthen research capacity and to enhance cross-fertilization of disciplines by recruiting new investigators and new expertise into the research community, and by developing and retaining these talents in a collaborative environment that fosters creativity and exploration.

We should think differently about how we organize ourselves around conducting RCTs. Clinical research should mirror clinical practice. We practice and should conduct clinical research within multidisciplinary teams: groups that share a collective mission designed to achieve shared goals. Leveraging “team” strengths (ie, respiratory therapist - nursing - medicine - pharmacist - physical therapist - nutritionist - economist team strengths) can unify us around clinical problems so that we can ask multiple primary and secondary questions within a single clinical trial. Most of us have witnessed stellar clinical outcomes when team members bring their best available disciplinary knowledge or disciplinary lens to the bedside. Similar results can occur when we join together to answer clinical research questions. This is not the traditional multidisciplinary framework where several disciplines work in parallel around a common problem, or the interdisciplinary framework where disciplines work jointly, sharing frameworks and methods. This new transdisciplinary framework may facilitate the creation of new knowledge using a new language and carries the potential for study findings to be enriched and be more rapidly translated into practice. Transdisciplinary frameworks will require full investigator partnership and meaningful methods to share effort, funding, and authorship.

For example, multidisciplinary innovation will continue to optimize the care provided to young patients with asthma. These patients require transitional care programs that span the continuum of care; they require novel interventions that bridge the work of bench and clinical scientists; they require economic modeling and re-evaluation; and they require both short-term and long-term outcome monitoring. Patients and families experience care that is tied to the collective output of all disciplines. Although we have made progress in conducting science in disciplinary silos, breakthrough science will require a transdisciplinary perspective.

Alternative Research Designs

Changing our approach to knowledge generation may require a paradigm shift in how we grade or stratify the quality of our research methods. Research methods are derived from the question posed. Understanding a patient's or family's perspective on a topic requires qualitative methodologies. Stellar examples of qualitative studies include Meert's program of research on end-of-life decision making.^35–37 Qualitative research often forms the basis for subsequent quantitative study, for example, when describing and then evaluating patient reports of dyspnea.

In addition, a well structured factorial design allows investigators to test 2 interventions with only a marginal increase in sample size.^27,30 The design offers a substantial benefit where multiple interventions need to be evaluated alone or in combination in the same population.²⁵ Such designs have been successfully used by the Acute Respiratory Distress Syndrome Network and offer an efficient solution when exploring interventions that may be interactive.

Bayesian statistical methods, which leverage the power of an adult trial to an associated pediatric trial, are being considered with increased frequency.^29,30 This approach links an a priori pediatric estimate of a treatment effect to previously or concurrently observed value in adults. Leveraging data from the adult trial substantially increases the power of the associated pediatric clinical trial. This statistical approach has the potential of making pediatric clinical trials feasible, because fewer pediatric subjects would be required to show efficacy.

Building Research Capacity

In the future we can expect more “virtual” research teams to come together to tackle clinical problems. Co-principal-investigators can sit on opposite sides of the country and the world. Free cyberspace links provide limitless opportunities to connect beyond the expensive and rudimentary conference call.

Highly functional research teams should not disband at the end of an RCT, but should launch directly into another project. Using existing structures will abbreviate start-up time and costs. Joining preexisting networks will strengthen our capacity to respond to evolving events such as the H1N1 flu epidemic. For example, coordinating the efforts of the PALISI network with other successful research groups, including the Canadian Critical Care Trials Group, the Australian and New Zealand Intensive Care Society Clinical Trials Group, and the European Society of Pediatric and Neonatal Intensive Care (to name a few), would leverage our capacity to respond rapidly to international issues.

We can build research capacity within the field by capitalizing on transdisciplinary and interdisciplinary mentoring. Clinical scientists need not practice independently. Finding mentors and colleagues and building a team with expertise in experimental design, protocol implementation, statistical analysis, scientific writing, research ethics, and grant preparation would booster our research capacity.

When choosing a science mentor, one should look for synergy.^38–40 You should have a real interest in your mentor's research, you should find their publications exciting, your mutual chemistry should feel right. Junior investigators need access to their mentors. Face time on a regular meeting schedule in person or electronically is important. Your mentor should help you launch your independent research career and provide you an opportunity to carve out your unique area of research. Stellar mentors are passionate about knowledge development. They are kind and truly interested in your work. They are generous and share their knowledge and experience. They give you their good ideas. They are inspiring, encouraging, humble, and your best advocate. Eventually you should be recognized as more expert than your mentor in your program of research.

Finally, finding the necessary funds to support clinical research continues to be challenging. Although “research” is often in the mission statement of many healthcare organizations, few are adequately structured to support clinical research. Indirect funds are often channeled to support the research enterprise that is dominated by laboratory-based research. Budgeting a clinical trial occurs concurrently with study design. Many professional organizations and philanthropic organizations have grant monies available for clinical projects, and some can be leveraged to support larger clinical trials.