Burden of disease

The burden of TBI for children is enormous—even when only considering children with the most severe injuries. In the 11 years from 1997 to 2007, the US CDC reports that 73,276 children died from TBI [1]. An analysis of the Kid’s Inpatient Database—utilizing discharge data from >3000 hospitals in 30 states—found more than 29,000 cases of hospitalized children with TBI, 4907 with severe TBI, a mortality rate of 24.2% [2] and $2.56 billion in acute hospital costs [3]. Internationally, the impact of TBI on child health is similarly large [4,5,6]. Our understanding of the long-term effect of severe TBI on children’s health is difficult to fully appreciate. One study estimates that at least 145,000 children were living with a TBI-related disability in 2005 [7] and the overall total life costs (medical costs and productivity losses) of injuries for children <14 years of age were $60.4 billion [8].

The paradox

The author Joseph Heller popularized the term “Catch-22,” broadly defined as a dilemma or difficult circumstance from which there is no escape because of mutually conflicting or dependent conditions. In our estimation, the field of severe TBI management appears to be in a similar situation at this time. Most clinicians would argue that a properly executed, randomized-controlled trial (RCT) would provide the best evidence for the efficacy of a proposed therapy. However, an important factor to the success of such an RCT would be to standardize as many of the clinical practices already in use except for the experimental therapy—as well as standardizing the populations of subjects under study based on patient and disease characteristics. Standardizing clinical practices across institutions would be most effective by applying strong recommendations from the literature that exists at the time of the planning of the RCT. The successful RCTs and rigorous guidelines would lead toward a consensus standard of care that could be altered as new information is generated. As outlined below, the interconnected circumstances of the field of pediatric neurotrauma care have hampered advances in the field up to this point.

Randomized controlled trials in children with severe TBI

Ten substantive randomized controlled trials [9,10,11,12,13,14,15,16,17,18] have been conducted in children with severe TBI (defined as age <18 years and Glasgow Coma Scale [GCS] score ≤8) over the past several decades, with 3 large, multi-centered studies completed in the last 10 years (Table 1). These 3 most recent trials all studied therapeutic hypothermia to improve overall outcome of the children and all demonstrated the difficulties in performing RCTs in the current environment. Hypothermia Pediatric Head injury Trial (Hyp-HIT) is the largest RCT conducted in children with severe TBI (n = 225) and represents a milestone achievement for the field [15]. The study was completed over 7 years in 17 clinical centers in Canada, Europe, and Australia and failed to find a beneficial effect of 24 h of therapeutic hypothermia. In performing the study, the experimental group had a greater incidence of hypotension and increased use of vasopressor support—both leading to lower cerebral perfusion pressure at critical times during the study. Moreover, hyperosmolar therapies were used much more commonly in the normothermia group while both children from both groups were exposed to hyperventilation (~40% of subjects). The Cool Kids Trial randomized children to receive 48 h of hypothermia or normothermia initially from 12 US sites which was expanded to sites within the USA, UK, and Australia after enrollment goals were not obtained [10]. The study was stopped for futility by the DSMB after approximately 3 years of enrollment and 77 patients randomized. As part of the study, the Executive Committee of Cool Kids proposed a treatment strategy for intracranial hypertension and other aspects of care based on the evidence-based guidelines to the study sites. However, several study sites expressed concerns that they did not routinely follow these practices related to glucose management, sedation practices, and other aspects of care. Ultimately, the clinical protocol became suggestions for sites to follow rather than mandates based on the sponsors’ desire to determine if hypothermia could be effective in the “real world.” Most recently, Beca and colleagues published a phase II RCT for hypothermia of slightly longer duration in 8 centers in Australia and New Zealand [11]. Over a 3.5-year period of enrollment, 764 children were screened that resulted in 50 evaluable children. All of these studies demonstrate the extreme difficulty in identifying patients in a timely enough manner to randomize children (generally within 6–8 h of injury), logistical problems of getting procedures implemented in sites located in a wide variety of countries and the need to screen many more patients to.

Table 1 Summary of randomized controlled trials for children with severe TBI
Full size table

Evidence-based guidelines

For children with severe TBI, evidence-based guidelines were first published in 2003 [19] and have been revised in 2012 [20]. For the most recent guidelines, an expert panel (15 clinicians including pediatric neurosurgeons, emergency medicine physicians, intensivists, anesthesiologists, neurologists and surgeons and 3 methodologists) was selected by the Brain Trauma Foundation based on their expertise. This panel determined topics for inclusion within the guidelines based on (i) the sufficiency of the evidence within the topic and (ii) the link between the topic and outcomes. Based on these criteria, 15 topics (Table 2) were selected including 8 medical interventions (hyperosmolar therapies, temperature, CSF diversion, barbiturates, hyperventilation, corticosteroids, analgesia/sedation/neuromuscular blockade, and nutrition/glucose). A doctoral-level librarian performed extensive literature searches to identify articles that met the inclusion criteria—clearly defined patient population of children with severe TBI, identifiable independent variables (treatments) and dependent variables (outcomes), adequate sample size—for each topic. Level I, II, and III recommendations were made for therapies that “must be done,” “should be considered,” and “may be considered,” respectively. The guidelines underwent peer review by 14 external reviewers and were reviewed and endorsed by 10 associations/societies including AAP—Section on Neurological Surgery, American Association of Neurological Surgeons, Society of Critical Care Medicine, Child Neurology Society, European Society of Pediatric and Neonatal Intensive Care and the Paediatric Intensive Care Society-UK.

Table 2 Summary of guidelines topics
Full size table

The new guidelines shed light on our inadequate knowledge of treatments for pediatric TBI. Specifically, there was insufficient evidence to support a Level I recommendation for any of the topics. Moreover, there was evidence to support only 4 Level II recommendations for medical therapies—(i) the use of corticosteroids is not recommended to improve outcome or reduce ICP, (ii) moderate hypothermia beginning early after severe TBI for only 24 h should be avoided, (iii) an immune-enhanced diet should be avoided, and (iv) HTS should be considered for treatment of intracranial hypertension. In summary, the existing literature cannot recommend that a clinician “must do” any aspect of the 15 therapies or maneuvers identified by the pediatric neurotrauma community. Importantly, there is only evidence that a clinician “should consider” use of HTS during intracranial hypertension episodes with the remaining 3 level II recommendations (hypothermia, steroids, immune-enhanced diets) suggesting that therapies should be avoided. Analysis of the level III recommendations offers clinicians little guidance as well (Table 3). As an example, the guidelines can only recommend that ICP monitoring “may be considered” and can only suggest that a threshold of 20 mmHg “may be considered”. However, many of the other recommendations are predicated on providing therapies during intracranial hypertension—which would not be diagnosed without the ICP monitor providing the necessary data and the clinician deciding on an appropriate threshold for ICP. Furthermore, basic questions regarding therapies that are widely believed to improve outcome and must be answered by the clinician caring for a child with severe TBI—Will CSF diversion lead to improved outcome? Are hyperosmolar therapies effective? Does prophylactic hyperventilation harm recovery? Should new methods to monitor for brain hypoxia be utilized? How many calories are needed for optimal recovery? When should glucose be administered?—remain unaddressed by the guidelines. This lack of evidence frustrates evidence-based clinical decision-making for all children with TBI and introduces uncontrollable variability into research protocols that attempt to standardize practices at multiple sites to successfully detect an experimental signal of a prospective therapy.

Table 3 Level III recommendations from the evidence-based guidelines
Full size table

Approaches and decisions for acute pediatric TBI (ADAPT) trial

The ADAPT Trial was designed to start to address the problems outlined above by understanding how therapies that are already in clinical use are associated with outcomes. Thus, ADAPT is an observational cohort study that intends to use statistical techniques common to comparative effectiveness research to develop new level II recommendations for the guidelines. The topics chosen included Tier 1 therapies for intracranial hypertension management (cerebrospinal fluid [CSF] diversion, hyperosmolar therapies, hyperventilation) and basic aspects of neurocritical care (hypoxia management [as measured by the partial pressure of brain oxygen (PbO2) monitoring], nutritional support and glucose management). For each topic, a primary and secondary hypothesis was generated and it is anticipated that ADAPT will generate up to 12 new Level II guidelines recommendations based on these a priori defined hypotheses. Importantly, these hypotheses will determine which of the currently used strategies within the topics are associated with improved neurological outcome (measured by the Glasgow Outcome Scale Extended for Pediatrics [GOS-E Peds] at 6 months after injury). If compelling enough, these level II recommendations may become new standards of care for years to come. As an example, if ADAPT demonstrates that initiation of nutritional support within a given number of days after TBI is associated with improved outcomes, this might be interpreted by the TBI community as sufficient evidence to adopt this approach as a standard for both clinical care and research protocols. Alternatively, findings from ADAPT could generate preliminary data that compels additional studies. In another example, PbO2 monitoring may be found to be associated with improved outcome after ADAPT is fully analyzed. In this instance, it might be necessary to perform further studies to develop a standard approach that is adopted by the pediatric neurotrauma community.

ADAPT was begun in July 2013 and steady progress has been made. Fifty-one clinical centers in 8 countries participated in patient enrollments, with enrollment of 1000 consecutive subjects completed in approximately 2.5 years. The last subjects were enrolled in September 2016 and primary outcomes were completed in late March 2017. Analysis of the primary hypotheses of the study will be completed after data cleaning and audits of all clinical sites are completed (anticipated for fall 2017). However, significant preliminary findings have already been published. In a survey of clinical sites participating in ADAPT, we found that significant variations of the medical goals of the clinical sites exist, let alone what actually gets accomplished in subjects within ADAPT [21]. In this analysis, we found that while sites generally targeted an ICP of 20 mmHg for their care, there were significant differences in strategies regarding all of the ADAPT hypotheses. We anticipate that completion of ADAPT will demonstrate even greater variation in clinical practices. We conducted a survey regarding EEG monitoring and treatment of seizures that demonstrated sites that utilize protocols report using fewer medications for treating electrographic abnormalities that are detected [22]. We have also published two analyses of subjects within the study—both analyses performed on the first 200 subjects enrolled in the study. In an analysis of the GCS scores that qualified subjects for ADAPT—documented by the providers who determined the need for ICP monitoring—validated that GCS scores incrementally predicted mortality in a tripartite distribution (GCS 3 vs. 4–5 vs. 6–8) [23]. Lastly, we demonstrated that variations in characteristics between children who had suffered from accidental and abusive head trauma [24].

Future plans

As stated previously, we anticipate publishing a number of reports regarding our primary hypotheses in the coming months that will expand the guidelines. We believe that these reports will serve as a basis for developing standards of care as well as new hypotheses to be tested. Given the large database, we also anticipate publishing findings on other aspects of the guidelines including mortality and disability prediction models, sedation practices, Tier 2 therapies (barbiturates, hypothermia, decompressive surgery for intracranial hypertenstion), fluid balance, ICP/CPP thresholds among others. We believe that understanding these findings—along with a better understanding of characterizing the disease of severe TBI—offers the best opportunity to make a difference in the field.