Signals of a silent death: direct and indirect evidence of bird-window collisions in a Neotropical university campus

Consistency between Direct and Indirect Trial Evidence: Is Direct Evidence Always More Reliable?

Bird collisions with buildings are an increasing concern and yet understanding the factors contributing to collisions at the species level remains largely unknown. This gap in our knowledge of species-specific strike patterns hinders the development of accurate estimates for the impact of death-by-collision on bird populations and impedes on our ability to minimize its effects. Our study offers the first examination of the impact of environmental variables on bird- window collisions at the species level. The Fatal Light Awareness Program Canada collected bird-window collision data in three distinct regions of Toronto, Canada during the migratory season of the years 2009 and 2010. Our results indicated that building percent window cover, exposed habitat cover, and cover of built structures significantly affect bird-window collisions. Multivariate analyses showed that the bird species that collided with buildings surrounded by a high level of urban greenery are species that typically occur in forested habitats and are foliage gleaners. In contrast, species that collided with buildings surrounded by a higher level of urbanization are species that typically occur in open woodland and are ground foragers. These results suggest that the composition of bird species colliding with buildings across various regions of the Greater Toronto Area is influenced by the local bird species community composition, by the configuration of the surrounding landscape, and by the levels of greenery around the buildings.

To evaluate two methods for assessing the changes in periodontitis grading in patients undergoing supportive periodontal therapy 10 years (T10) after retrospective baseline grading. The periodontitis grade of 51 supportive periodontal therapy patients was assessed using indirect evidence as the primary criterion for periodontitis progression at baseline and T10 (radiographic bone loss/age index, periodontitis phenotype). Grading at T10 was also performed using the direct evidence for periodontitis progression (clinical attachment loss over the previous 5 years). The use of indirect evidence for periodontal progression at baseline and T10 was defined as method 1 to assess the changes in periodontitis grading. The use of indirect evidence at baseline and direct evidence at T10 was defined as method 2. Changes in periodontitis grading using methods 1 and 2 were evaluated (Wilcoxon signed-rank test). Agreement between methods 1 and 2 was assessed (Cohen kappa). Indirect baseline grading revealed five grade B and 46 grade C patients. The indirect grading at T10 revealed 17 grade B and 34 grade C patients. The direct T10-grading classified all patients as grade C. Method 1 led to an overall improvement in periodontitis grading after 10 years of supportive periodontal therapy (P = .0030), whereas method 2 led to a deterioration (P = .0369). The comparison between methods 1 and 2 showed that they led to different results in terms of grading (Cohen kappa = 0.116208). Periodontitis grading may change during supportive periodontal therapy. Using indirect or direct evidence as the primary grading criterion during supportive periodontal therapy may lead to different results.

BackgroundAdaptive platform trials allow randomized controlled comparisons of multiple treatments using a common infrastructure and the flexibility to adapt key design features during the study. Nonetheless, they have been criticized due to the potential for time trends in the underlying risk level of the population. Such time trends lead to confounding between design features and risk level, which may introduce bias favoring one or more treatments. This is particularly true when experimental treatments are not all randomized during the same time period as the control, leading to the potential for bias from non-concurrent controls.MethodsTwo analysis methods addressing this bias are stratification and adjustment. Stratification uses only comparisons between treatment cohorts randomized during identical time periods and does not use non-concurrent randomizations. Adjustment uses a modeled analysis including time period adjustment, allowing all data to be used, even from periods without concurrent randomization. We show that these competing approaches may be embedded in a common framework using network meta-analysis principles. We interpret the stages between adaptations in a platform trial as separate fixed design trials. This allows platform trials to be viewed as networks of direct randomized comparisons and indirect non-randomized comparisons. Network meta-analysis methodology can be re-purposed to aggregate the total information from a platform trial and to transparently decompose this total information into direct randomized evidence and indirect non-randomized evidence. This allows sensitivity to indirect information to be assessed and the two analysis methods to be clearly compared.ResultsSimulations of platform trials were analyzed using a network approach implemented in the netmeta package in R. The results demonstrated bias of unadjusted methods in the presence of time trends in risk level. Adjustment and stratification were both unbiased when direct evidence and indirect evidence were consistent. Network tests of inconsistency may be used to diagnose inconsistency when it exists. In an illustrative network analysis of one of the treatment comparisons from the STAMPEDE platform trial in metastatic prostate cancer, indirect comparisons using non-concurrent controls were inconsistent with the information from direct randomized comparisons. This supports the primary analysis approach of STAMPEDE, which used only direct randomized comparisons.ConclusionNetwork meta-analysis provides a natural methodology for analyzing the network of direct and indirect treatment comparisons from a platform trial. Such analyses provide transparent separation of direct and indirect evidence, allowing assessment of the impact of non-concurrent controls. We recommend time-stratified analysis of concurrently controlled comparisons for primary analyses, with time-adjusted analyses incorporating non-concurrent controls reserved for secondary analyses. However, regardless of which methodology is used, a network analysis provides a useful supplement to the primary analysis.

ObjectiveTo investigate the capability of periodontal grading to estimate the progression of periodontal disease and the responsiveness to therapy.Materials and methodsEighty-four patients who underwent non-surgical therapy (NST) were included. Direct and indirect evidence of progression were determined according to the current classification. Responsiveness to therapy was examined using mean pocket probing depths reduction (PPDRed), reduction of bleeding on probing (BOPRed), and the rate of pocket closure (%PC) after six months.ResultsStatistical analysis revealed no agreement between direct and indirect evidence in grading periodontitis (κ = 0.070). The actual rate of progression as determined by longitudinal data was underestimated in 13% (n = 11), overestimated in 51% (n = 43) and correctly estimated in 30% (n = 36) by indirect evidence. No significant differences in responsiveness to therapy were observed in patients graded according to direct evidence. Using indirect evidence, patients assigned grade C showed more PPDRed but less BOPRed and lower %PC compared to grade B.ConclusionThe present data indicate that indirect evidence may lead to inaccuracies compared to direct evidence regarding the estimation of periodontal progression. However, indirect evidence seems to be more suitable in the estimation of responsiveness to therapy than direct evidence, helping to identify cases that are more likely to require additional therapies such as re-instrumentation or periodontal surgery.Clinical relevanceRegarding the estimation of disease progression and responsiveness to periodontal therapy, accuracy and reliability of both direct and indirect evidence are limited when grading periodontitis.

Bird-window collisions are a dramatic cause of bird mortality globally. In Latin America, statistics are generally very scarce and/or inaccessible so the frequency of such incidents is still poorly understood. Nevertheless, civilians have applied preventive methods (e.g. adhesive bird-of-prey decals) sparsely but, to our knowledge, no study has evaluated their effectiveness in Brazil. Here, we estimated the mortality rate of bird-window collisions and tested the effectiveness of bird-of-prey decals at preventing such accidents. We undertook daily searches for bird carcasses, presumably resulting from window collisions, near all buildings on a university campus over seven months. Adhesive bird-of-prey decals were then applied to the two buildings with the highest mortality rates and surveys continued for over 12 more months. The mortality rates before and after the application of decals and between seasons were then compared using Friedman test. We recorded 36 collisions, 29 around the two buildings with the highest collision rates 19 prior and 10 after our intervention with associated collision rates of 0.08 and 0.04 collisions/day. Although mortality was reduced by almost half, this difference was not statistically significant. The Blue-black grassquit, Volatinia jacarina (Linnaeus, 1766), and Ruddy ground dove, Columbina talpacoti (Temminck, 1810) suffered the highest number of collisions, followed by the Rufous-collared sparrow, Zonotrichia capensis (P. L. Statius Muller, 1776). Our bird-of-prey decals and efforts were insufficient to prevent or dramatically reduce the number of bird-window collisions. Therefore, we recommend that different interventions be used and additional long-term studies undertaken on their efficacy.

Sometimes direct evidence is so strong that a prescription for practice is decreed. Usually, things are not that simple—leaving aside the possibility that important trade offs may be involved, direct...

Familiar statistical tests and estimates are obtained by the direct observation of cases of interest: a clinical trial of a new drug, for instance, will compare the drug's effects on a relevant set of patients and controls. Sometimes, though, indirect evidence may be temptingly available, perhaps the results of previous trials on closely related drugs. Very roughly speaking, the difference between direct and indirect statistical evidence marks the boundary between frequentist and Bayesian thinking. Twentieth-century statistical practice focused heavily on direct evidence, on the grounds of superior objectivity. Now, however, new scientific devices such as microarrays routinely produce enormous data sets involving thousands of related situations, where indirect evidence seems too important to ignore. Empirical Bayes methodology offers an attractive direct/indirect compromise. There is already some evidence of a shift toward a less rigid standard of statistical objectivity that allows better use of indirect evidence. This article is basically the text of a recent talk featuring some examples from current practice, with a little bit of futuristic speculation.

Pooling of direct and indirect evidence from randomized trials, known as mixed treatment comparisons (MTC), is becoming increasingly common in the clinical literature. MTC allows coherent judgements on which of the several treatments is the most effective and produces estimates of the relative effects of each treatment compared with every other treatment in a network.We introduce two methods for checking consistency of direct and indirect evidence. The first method (back-calculation) infers the contribution of indirect evidence from the direct evidence and the output of an MTC analysis and is useful when the only available data consist of pooled summaries of the pairwise contrasts. The second more general, but computationally intensive, method is based on 'node-splitting' which separates evidence on a particular comparison (node) into 'direct' and 'indirect' and can be applied to networks where trial-level data are available. Methods are illustrated with examples from the literature. We take a hierarchical Bayesian approach to MTC implemented using WinBUGS and R.We show that both methods are useful in identifying potential inconsistencies in different types of network and that they illustrate how the direct and indirect evidence combine to produce the posterior MTC estimates of relative treatment effects. This allows users to understand how MTC synthesis is pooling the data, and what is 'driving' the final estimates.We end with some considerations on the modelling assumptions being made, the problems with the extension of the back-calculation method to trial-level data and discuss our methods in the context of the existing literature.

If a mother contracts the Zika Virus before or during pregnancy, then there is a risk of the child developing Congenital Zika Syndrome (CZS). An infant can then experience problems feeding due to the specific physical and developmental consequences of Congenital Zika Syndrome (CZS), such as microcephaly, dysphagia and an increased likelihood of choking. This qualitative evidence synthesis accesses direct and indirect evidence to inform WHO infant feeding guidelines. We conducted a qualitative evidence synthesis of the values and preferences of relevant stakeholders (e.g. pregnant women, mothers, family members and health practitioners) concerning infant (0–2 years) feeding in the presence of: 1) CZS (the‘direct evidence’); 2) severe disability and nonprogressive, chronic encephalopathies (‘indirect evidence’), which present with similar problems. Authors’ findings were extracted, synthesised using thematic synthesis techniques, and confidence in the findings were assessed using GRADE-CERQual. Six CZS-specific studies (all from Brazil) were included in the direct evidence, with a further eight indirect studies reporting feeding difficulties in infants with severe disability and nonprogressive, chronic encephalopathies. Included studies highlighted: breast-feeding represented the preference for all mothers in the studies in both reviews, and the inability to do so affected bonding between parents and child, and generated fear and anxiety relating to feeding choices, especially around the risks of choking and swallowing; the perception that health professionals were often unable to offer appropriate advice; the potential value of training; and a strong desire to achieve individual maternal autonomy in infant feeding decisions. Confidence in most findings ranged from low to moderate. The evidence base has limitations, but consistently reported that parents of children with feeding difficulties due to Congenital Zika Syndrome, or similar, need information, advice and counselling, and substantial emotional support. Parents perceive that these needs are often neither recognised nor satisfied; optimal feeding and support strategies for this population have not yet been identified.

If a mother contracts the Zika Virus before or during pregnancy, then there is a risk of the child developing Congenital Zika Syndrome (CZS). An infant can then experience problems feeding due to the specific physical and developmental consequences of Congenital Zika Syndrome (CZS), such as microcephaly, dysphagia and an increased likelihood of choking. This qualitative evidence synthesis accesses direct and indirect evidence to inform WHO infant feeding guidelines. We conducted a qualitative evidence synthesis of the values and preferences of relevant stakeholders (e.g. pregnant women, mothers, family members and health practitioners) concerning infant (0-2 years) feeding in the presence of: 1) CZS (the'direct evidence'); 2) severe disability and nonprogressive, chronic encephalopathies ('indirect evidence'), which present with similar problems. Authors' findings were extracted, synthesised using thematic synthesis techniques, and confidence in the findings were assessed using GRADE-CERQual. Six CZS-specific studies (all from Brazil) were included in the direct evidence, with a further eight indirect studies reporting feeding difficulties in infants with severe disability and nonprogressive, chronic encephalopathies. Included studies highlighted: breast-feeding represented the preference for all mothers in the studies in both reviews, and the inability to do so affected bonding between parents and child, and generated fear and anxiety relating to feeding choices, especially around the risks of choking and swallowing; the perception that health professionals were often unable to offer appropriate advice; the potential value of training; and a strong desire to achieve individual maternal autonomy in infant feeding decisions. Confidence in most findings ranged from low to moderate. The evidence base has limitations, but consistently reported that parents of children with feeding difficulties due to Congenital Zika Syndrome, or similar, need information, advice and counselling, and substantial emotional support. Parents perceive that these needs are often neither recognised nor satisfied; optimal feeding and support strategies for this population have not yet been identified.

Severe Traumatic Brain Injury in Infants, Children, and Adolescents in 2019: Some Overdue Progress, Many Remaining Questions, and Exciting Ongoing Work in the Field of Traumatic Brain Injury Research In this Supplement to Pediatric Critical Care Medicine, we are pleased to present the Third Edition of the Guidelines for the Management of Pediatric Severe Traumatic Brain Injury (TBI). This body of work updates the Second Edition of the guidelines that was published in 2012 (1). It represents a substantial effort by a multidisciplinary group of individuals assembled to reflect the team approach to the treatment of these complex, critically ill patients that is essential to optimizing critical care and improving outcomes. This work also represents the strong and always-evolving partnership between investigators from the medical and research communities, forged in Chicago in 2000, from which the first pediatric TBI guidelines were developed. The mutual trust and respect we share have been the foundation of our commitment to bringing evidence-based care to children with TBI. Updating these guidelines was particularly exciting to the individuals who have participated in the previous two editions because several new studies have been published which begin to address a number of major gaps in the pediatric TBI literature—gaps that were specifically identified as targets for future research in earlier editions. For example, we are now able to include reports on the effects of commonly used sedatives and analgesics on intracranial pressure (ICP). Similarly, initial head-to-head comparisons of the influence of agents in routine "real world" use such as hypertonic saline (HTS), fentanyl, and others now inform these guidelines (2,3). A total of 48 new studies were included in this Third Edition. Although some progress has been made and should be celebrated, overall the level of evidence informing these guidelines remains low. High-quality randomized studies that could support level I recommendations remain absent; the available evidence produced only three level II recommendations, whereas most recommendations are level III, supported by low-quality evidence. Based in part on a number of requests from the readership to individual clinical investigators, we have included a companion article in the regular pages of Pediatric Critical Care Medicine that presents a "Critical Pathway" algorithm of care for both first-tier and second-tier (refractory intracranial hypertension) approaches. The algorithm reflects both the evidence-based recommendations from these guidelines and consensus-based expert opinion, vetted by the clinical investigators, where evidence was not available. An algorithm was provided in the First but not Second Editions of the guidelines, and we believe that given the new reports available, along with the existing gaps in evidence, a combination of evidence-based and consensus-based recommendations provides additional and much-needed guidance for clinicians at the bedside. The algorithm also addresses a number of issues that are important but were not previously covered in the guidelines, given the lack of research and the focus on evidence-based recommendations. This includes addressing issues such as a stepwise approach to elevated ICP, differences in tempo of therapy in different types of patients, scenarios with a rapidly escalating need for ICP-directed therapy in the setting of impending herniation, integration of multiple monitoring targets, and other complex issues such as minimal versus optimal therapeutic targets and approaches to weaning therapies. We hope that the readership finds the algorithm document helpful, recognizing that it represents a challenging albeit important step. Designing and developing this pediatric TBI evidence-based guidelines document required an expert administrative management team, and to that end, we are extremely grateful to the staff of the Pacific Northwest Evidence-based Practice Center, Oregon Health & Science University, for their vital contribution to this work. We are also grateful to the Brain Trauma Foundation and the Department of Defense for supporting the development and publication of these guidelines documents. We are grateful to the endorsing societies for recognizing the importance of this work and for the considerable work of the clinical investigators in constructing the final document. We are also pleased to have collaborated with the Congress of Neurological Surgeons and the journal Neurosurgery that is copublishing the Executive Summary document of these guidelines for its readership. We are also grateful to Hector Wong for serving as Guest Editor, along with the external reviewers of this final document. Finally, we thank each of the clinical investigators and coauthors on this project. We believe that the considerable uncompensated time and effort devoted to this important project will help to educate clinicians worldwide and enhance the outcomes of children with severe TBI. Clinical investigators provided Conflict of Interest Disclosures at the beginning of the process, which were re-reviewed at the time of publication. No clinical investigator made inclusion decisions or provided assessments on publications for which they were an author. Looking forward, it is important to recognize that these guidelines were written as the Approaches and Decisions in Acute Pediatric TBI Trial (ADAPT) (4–6), one of the most important in the field of pediatric TBI, was coming to a close. The ADAPT completed enrollment of 1,000 cases of severe pediatric TBI and is one example of the recent heightened general interest in TBI as a disease. This new interest in the importance of TBI has emerged in part from the recognition of the high prevalence of TBI across the injury severity spectrum, particularly concussion, and from the need for new classification systems and new trial design for TBI in both children and adults (7,8). In addition, the emerging links between TBI and a number of neurodegenerative diseases have broadened the interest in TBI, have led to additional support of TBI research, and have produced an unprecedented level of research in TBI and a quest for new therapies (9–11). We expect that the results of ADAPT, along with those of other ongoing and recently completed research in the field, will help provide new insight and clarity into the acute medical management (MM) of infants, children, and adolescents with severe TBI, and mandate further refinement of the recommendations in these documents. We know that we speak for the entire team of clinical investigators in welcoming the opportunity to incorporate additional high-level evidence into future updates of these guidelines. METHODS The methods for developing these guidelines were organized in two phases: a systematic review, assessment, and synthesis of the literature; and use of that product as the foundation for evidence-based recommendations. These guidelines are the product of the two-phased, evidence-based process. Based on almost 2 decades of collaboration, the team of clinical investigators and methodologists (Appendix A, Supplemental Digital Content 1, https://links.lww.com/PCC/A774) is grounded in and adheres to the fundamental principles of evidence-based medicine to derive recommendations, and is committed to maintaining the distinction between evidence and consensus. It is important that this distinction is clear to promote transparency and inspire innovative future research that will expand the evidence base for TBI care. Because these guidelines only provide recommendations based on available evidence, most often they do not provide direction for all phases of clinical care. Ideally, clinically useful protocols begin with evidence-based guidelines, and then use clinical experience and consensus to fill the gaps where evidence is insufficient. The goal is to use the evidence and the evidence-based recommendations as the backbone to which expertise and consensus can be added to produce protocols appropriate to specific clinical environments (Fig. 1, "Future Research section"). In a process independent from developing this Third Edition of the guidelines, the team engaged in a consensus process and produced the algorithm for treatment of severe TBI in pediatric patients.Figure 1.: Dynamic process for guidelines, protocols, and future research. The diagram shows the flow of information from available evidence to a guideline. The guideline leads to gaps that identify future research and consensus-based clinical protocols that fill gaps, both of which lead to a generation of new research.The following "Methods section" describes the process we used to produce the systematic review and evidence-based recommendations. The methods used to develop the algorithm are described in that document (12). Phase I: Systematic Evidence Review and Synthesis Scope of the Systematic Review Criteria for Including Publications Appendix B (Supplemental Digital Content 1, https://links.lww.com/PCC/A774) lists the criteria for including studies for review using the categories of population, interventions, comparators, outcomes, timing, settings, study designs, and publication types. The criteria for population are as follows: Age 18 years old or younger TBI Glasgow Coma Scale (GCS) score less than 9 Included Topics. The team chose to carry forward topics from the Second Edition of these guidelines. No new topics were added. The topics are organized in three categories that are specific to severe TBI in children: monitoring, thresholds, and treatments. Monitoring 1. ICP 2. Advanced neuromonitoring 3. Neuroimaging Thresholds 4. ICP 5. Cerebral perfusion pressure (CPP) Treatments 6. Hyperosmolar therapy 7. Analgesics, sedatives, and neuromuscular blockade (NMB) 8. Cerebrospinal fluid (CSF) drainage 9. Seizure prophylaxis 10. Ventilation therapies 11. Temperature control 12. Barbiturates 13. Decompressive craniectomy 14. Nutrition 15. Corticosteroids Major Changes for This Edition. Major changes for this edition are summarized here, and details are provided in Appendix C (Supplemental Digital Content 1, https://links.lww.com/PCC/A774). The clinical investigators and methods team identified three primary endpoints considered important health outcomes for pediatric patients with TBI: To improve overall outcomes (mortality, morbidity, function) To control ICP To prevent posttraumatic seizures (PTSs) Two new meta-analyses were added to the evidence base for temperature control. The title of "Hyperventilation" was changed to "Ventilation Therapies." Recommendations are provided as level I, II, or III. In some cases, publications from the second edition were not included in this 3rd Edition. Our rationale for excluding previously included studies was based on identification of current material that superseded our earlier work (See Appendix E, Supplemental Digital Content 1, https://links.lww.com/PCC/A774). Similarly, we removed or changed recommendations from the 2nd Edition when the current literature provided new and/or more accurate information (see Appendix A, Supplemental Digital Content 1, https://links.lww.com/PCC/A774). Study Selection and Compilation of Evidence Literature Search Strategies. The research librarian who worked on the Second Edition reviewed and updated the search strategies for that edition and executed the searches for this Third Edition. Ovid/MEDLINE was searched from 2010 to May of 2015, and an update was performed to include articles published and indexed through June of 2017. Publications recommended by peers that were not captured in the search were reviewed, and those meeting inclusion criteria were included in the final library. The search strategy is in Appendix D (Supplemental Digital Content 1, https://links.lww.com/PCC/A774). Abstract and Full-Text Review. Abstracts for publications captured in the search were reviewed independently by two members of the methods team. Articles were retained for full-text review if at least one person considered them relevant based on the abstract. Two methods team members read each full-text article and determined whether it met the inclusion criteria (Appendix B, Supplemental Digital Content 1, https://links.lww.com/PCC/A774). The included and excluded full-text articles for each topic were also reviewed by one or more clinical investigators who took the lead on each topic, and full-text articles were available for review by all authors. The key criteria for inclusion were as follows: the study population was pediatric patients (age, ≤ 18 yr old) with severe TBI (defined as GCS score of 3–8) and the study assessed an included outcome. Publications with samples that included adults, moderate or mild severities, or pathologies other than TBI (indirect evidence) were considered when direct evidence was limited or not available. Discrepancies between reviewers were resolved via consensus or by a third reviewer. A list of studies excluded after full-text review is in Appendix E (Supplemental Digital Content 1, https://links.lww.com/PCC/A774). Use of Indirect Evidence and Intermediate Outcomes Direct evidence comes from studies that compare important health outcomes (e.g., mortality, morbidity, function) between two or more intervention groups or between an intervention group and a control group that represent the population of interest, in this case pediatric patients with severe TBI. When direct evidence was limited or not available, indirect evidence was used to support a recommendation. Indirect evidence has been defined in previous work by this methods team (1,13,14) and other evidence-based methods groups (15,16). In this edition, we included two types of indirect evidence. 1. Evidence That Improvement in an Intermediate Outcome Is Associated With Important Health Outcomes In some cases, there is a lack of direct evidence that utilization of a specific treatment option results in improved patient outcomes such as mortality or morbidity, but there is evidence about changes in an intermediate outcome, which is then associated with improved mortality or morbidity. The most notable intermediate outcome for the treatment of TBI is management of ICP. Multiple studies (cited in the ICP Monitoring topic of this guideline) consistently demonstrate that patients whose ICP is successfully maintained at or under a maximum threshold have reduced mortality and improved function. As a consequence, the clinical investigators elected to identify "Control of ICP" as an important intermediate outcome, and use the available indirect evidence to support the recommendations about monitoring ICP and for treatments designed to lower ICP. Intermediate outcomes and indirect evidence of this nature were used in three topics for this edition of the guidelines: ICP Monitoring, Ventilation Therapies, and Temperature Control. In each of these topics, an intermediate outcome was used as the endpoint because, although direct evidence was lacking that intervening improves mortality or function, indirect evidence was available associating management of the intermediate outcome with improved mortality or function. For ICP monitoring, the intermediate outcome was managed ICP; indirect evidence that patients with managed ICP had better outcomes was used to support the recommendation. For ventilation therapies, the intermediate outcomes were prevention of severe hypocarbia (SH). There were no pediatric studies that directly related hyperventilation to poor outcomes. However, there was evidence of an association between SH and mortality; thus, studies that demonstrated this association were used as indirect evidence. For temperature control, the intermediate outcomes were mean and peak CSF myelin basic protein concentrations and phenytoin levels. 2. Evidence From Samples With Mixed Ages, Severities, or Pathologies In some cases, when direct evidence was lacking, we considered studies that included patients with mixed severities (mild, moderate, and severe TBI), mixed ages, or mixed pathologies (traumatic and non-TBI) using the following criteria: How relevant to (or different from) our target population is the population in the indirect study? To what extent does the relevant physiology of the population in the indirect study approximate the relevant physiology of the population of interest? To what extent are differences in physiology expected to influence the outcome? In what direction would these differences influence the observed effect? In this edition, indirect evidence from studies with mixed severities, ages, or pathologies was included in the topics about analgesics, sedatives, and NMB; CSF drainage; and seizure prophylaxis. When indirect evidence was included, it is noted in the table describing the quality of the body of evidence. Quality Assessment of Individual Studies All included studies were assessed for potential for bias, which is an approach to assessing the internal validity or quality of an individual study. This assessment is a core component of systematic review methods. It is an approach to considering and rating studies in terms of how the study design and conduct addressed issues such as selection bias, confounding, and attrition. The criteria used for this edition are described in Appendix F (Supplemental Digital Content 1, https://links.lww.com/PCC/A774). Two reviewers independently evaluated each study using the criteria appropriate for the study design (i.e., randomized controlled trials [RCTs], observational studies, studies of thresholds) and rated the study as class 1, 2, or 3 evidence based on the combination of study design and conduct. Class 1 is the highest class and is limited to good-quality RCTs. Class 2 includes moderate-quality RCTs and good-quality cohort or case-control studies. Class 3 is the lowest class and is given to low-quality RCTs, moderate- to low-quality cohort or case-control studies, and treatment series and other noncomparative designs. Differences in ratings were reconciled via consensus or the inclusion of a third reviewer as needed. Data Abstraction Data were abstracted from studies by a member of the methods team and checked for accuracy by a second member. Information was recorded about the study population, design, and results. Key elements of each included study are presented in the Summary of Evidence tables for each topic. Complete abstraction tables are available upon request. Synthesis The final phase of the evidence review is the synthesis of individual studies into information that the clinical investigators and the methods team use to develop recommendations. This synthesis is described for each topic in the section titled "Evaluation of the Evidence," following the Recommendations and preceding the Evidence Summary. Identification of Subtopics and Synthesis For each monitoring, thresholds, or treatment topic, the clinical investigators identified important subtopics or clinical questions. The studies in each topic were reviewed to determine if quantitative synthesis—meta-analysis—was feasible. This involved determining if the patient populations, specifics of the intervention, and the outcomes were similar enough across several studies that the study results could be combined. The result of this assessment is included in the Quality of the Body of Evidence table for each subtopic. For this edition, we did not identify any topics for which quantitative synthesis was appropriate according to current standards. For this reason, the evidence was synthesized qualitatively. Quality of the Body of Evidence Assessing the quality of the body of evidence involves four domains: the aggregate quality of the included individual studies, the consistency of the results across studies, whether the evidence provided is direct or indirect, and the precision of the estimates of the outcomes. The criteria and ratings are outlined below, and more detailed definitions are given in Appendix G (Supplemental Digital Content 1, https://links.lww.com/PCC/A774). In addition, the number of studies and number of included subjects are considered. Based on these, an overall assessment is made as to whether the quality of the body of evidence is high, moderate, low, or insufficient. The assessment of the body of evidence for each subtopic is included in a summary table in each section following the recommendations. Criteria Quality of Individual Studies: This identifies the quality of the individual studies. It details how many studies are class 1, class 2, and class 3. Consistency: is the extent to which the results and are similar across studies. It is rated high are moderate are or one is more It is not when the body of evidence of a study. We as whether the study population is the as the population of interest and whether the outcomes are clinical than intermediate outcomes. Evidence is as indirect, or is the of the for a given outcome. is rated high, moderate, or low. How this is determined on the of used in a specific study but include of the of other of or the of used to determine These criteria are then considered when a rating to the body of evidence. The ratings are defined as follows: that the evidence reflects the research is to the in the of that the evidence reflects the research the in the of and the that the evidence reflects the research is to the in the of and is to the Evidence is or does not a A of quality of the body of evidence a about the importance of the and these across topics and The following general are provided to the that are but are not as two or more class 1 studies demonstrate for a topic, the overall quality of the body of evidence be assessed as because there is about the Similarly, class 1 or 2 studies that provide indirect evidence only low-quality evidence In some cases, the body of evidence be a but the rating A study a body of evidence if it is a class 1 a moderate-quality body of evidence if it is a class 2 study with a and moderate or evidence if the is and the precision of the of is low. is the extent to which research are useful for informing recommendations for a population the population that is the target of the is important to when assessing will on the topic, and the assessment is there is no rating for focus on the of the patient population (e.g., to which patients are the results and the for care (e.g., where could a similar result be if the patient population the inclusion criteria for the review, there be specific that The of the setting in which a study was also be important to For example, a study in a or not be to other settings, on how similar the are to the population of interest or how similar the of the is to the care setting of to be considered include the (e.g., or and the of (e.g., level of The and of are considered because it is that the patients, and available are different across In this edition, we the of individual studies in the of the Body of Evidence and section" following the recommendations. Phase of Recommendations of Recommendations Class 1, 2, or 3 studies the evidence on which the recommendations are our current identification of evidence is but not for the development of recommendations. No recommendations were made a in evidence. evidence was whether it could be used to inform recommendations was based on the quality of the body of evidence and of there were cases in which evidence was but the quality was and our to the evidence into recommendations. if a was not the evidence was included for future because in the new studies be in changes in the assessment of the quality of the body of evidence. of Recommendations in this edition are as level I, level II, or level III. The level of is determined by the assessment of the quality of the body of evidence, than the class of the included studies. The were based on the quality of the body of evidence as follows: I recommendations were based on a body of evidence. II recommendations were based on a moderate-quality body of evidence. recommendations were based on a low-quality body of evidence. could result in a level (e.g., a body of with In this edition, was not used to a recommendation. However, given the lack of and methods in this we issues that were identified and by the clinical was used in cases where there were no studies identified or because the body of evidence had major quality the evidence was no recommendations were Review and of the literature review, identification of new studies, quality assessment, and the methods team for each topic to two clinical The clinical investigators read the included studies and the recommendations, provided and additional studies for team members the and reviewed new studies, and provided the clinical investigators with new publications and a summary of the evidence for each topic. Clinical Review In a meeting in each topic was presented and by the Based on these the methods team the guidelines. Review of Complete The of all topics and the other of the guidelines (e.g., Supplemental Digital Content 1, https://links.lww.com/PCC/A774) was to all clinical investigators for review and and through to the and the document. Review. were made based on from the clinical investigators, the Third Edition and an Executive Summary were to the journal Pediatric Critical Care Medicine for A review was also by members of the of Neurological of Neurological Surgeons Guidelines Review in with the clinical investigators and methods team, to publication in the journal ICP Monitoring Recommendations of I and II There was evidence to support a level I or II for this topic. To Use of ICP monitoring is Changes From Edition. There are no changes from the Second Edition to the recommendations. new class 3 observational studies were added to the evidence base for this topic injury to the after severe TBI is a result of a of that perfusion of and and of and Brain from and or the of the leads to intracranial further and of ICP represents a key in the of injury phase following TBI the in both and outcome after severe TBI have been using

A review of the evidence concerning the effect of chronic or intermittent hypoxia on cognition in childhood was performed by using both a systematic review of the literature and critical appraisal criteria of causality. Because of the significant impact of behavioral disorders such as attention-deficit/hyperactivity disorder on certain cognitive functions as well as academic achievement, the review also included articles that addressed behavioral outcomes. Both direct and indirect evidence were collected. A structured Medline search was conducted from the years 1966-2000 by using the OVID interface. Both English- and non-English-language citations were included. Significant articles identified by the reviewers up to 2003 were also included. To be included as direct evidence, an article needed to be an original report in a peer-reviewed journal with data on cognitive, behavioral, or academic outcomes in children up to 14 years old, with clinical conditions likely to be associated with exposure to chronic or intermittent hypoxia. Indirect evidence from other reviews and publications in closely related fields, including experimental studies in adults, was used to help formulate conclusions. Two reviewers screened abstracts and titles. Each article included as direct evidence received a structured evaluation by 2 reviewers. Adjudication of differences was performed by a group of 2 reviewers and a research consultant. After this review, tables of evidence were constructed that were used as the basis for group discussion and consensus development. Indirect evidence assigned by topic to specific reviewers was also presented as part of this process. A formal procedure was used to rank the studies by design strength. The critical appraisal criteria for causation described in Evidence Based Pediatrics and Child Health (Moyer V, Elliott E, Davis R, et al, eds. London, United Kingdom: BMJ Books; 2000:46-55) were used to develop consensus on causality. A total of 788 literature citations were screened. For the final analysis, 55 articles met the criteria for inclusion in the direct evidence. Of these, 43 (78.2%) reported an adverse effect. Of the 37 controlled studies, 31 (83.8%) reported an adverse effect. Adverse effects were noted at every level of arterial oxygen saturation and for exposure at every age level except for premature newborns. The studies were classified into 5 clinical categories: congenital heart disease (CHD), sleep-disordered breathing (SDB), asthma, chronic ventilatory impairment, and respiratory instability in infants. Two of these categories, CHD and SDB, which accounted for 42 (76.4%) of the included articles, fulfilled the Evidence Based Pediatrics and Child Health criteria for causation. The indirect evidence included 8 reviews, 1 meta-analysis, and 10 original reports covering the fields of adult anoxia, animal research, SDB in adults, natural and experimental high-altitude studies, perinatal hypoxic-ischemic encephalopathy, anemia, and carbon-monoxide poisoning. The studies of high-altitude and carbon-monoxide poisoning provided evidence for causality. Adverse impacts of chronic or intermittent hypoxia on development, behavior, and academic achievement have been reported in many well-designed and controlled studies in children with CHD and SDB as well as in a variety of experimental studies in adults. This should be taken into account in any situation that may expose children to hypoxia. Because adverse effects have been noted at even mild levels of oxygen desaturation, future research should include precisely defined data on exposure to all levels of desaturation.

Although dinoprostone is widely used for cervical ripening, there are other methods available that do not need cold storage, might be less expensive and might be equally or more effective. These include Foley catheter and misoprostol. Many randomised controlled trials have compared these different interventions, but no large single trial has compared them all head to head. A systematic review of all of the randomised trials making comparisons of these interventions has been undertaken (Chen et al., BJOG 2015; DOI: 10.1111/1471-0528.13456). Conventio-nal meta-analysis can estimate the comparative effectiveness of each pair of treatments that have been compared in a trial. However, such direct evidence is not available for all possible pairwise combinations of treatments. Even if it were, it still would not be possible to say which treatment was the most effective without making informal indirect comparisons between these pooled pairwise estimates. In the absence of large randomised controlled trials comparing all treatments, a joint analysis of all of the relevant evidence is required to determine which of the possible interventions is most effective. A network meta-analysis is a statistical method for estimating the comparative effects of all treatments in a network using all trials comparing any combination of these treatments (Mills et al. BMJ 2013,346:f2914). This allows us to use both direct and indirect evidence and to assess its consistency. If direct evidence is available then direct and indirect estimates can be compared and jointly synthesised to provide a coherent set of estimates for all three treatments; and the probability that each treatment is the most effective (or second most effective etc.) can be calculated. As with any statistical method, there are assumptions. Network meta-analysis requires all the assumptions of standard pairwise meta-analyses. However, one further assumption called the similarity assumption is also required. There are several ways to conceptualise this but in short it assumes that the trials do not differ in the distribution of effect modifiers between different pairwise comparisons (Salanti, Res Synth Meth 2012;3:80–97). For example, if Foley catheter was more effective in older women then the distribution of age across trials comparing Foley catheter with each other treatment should be about the same. There are a number of methods for statistically checking this assumption by comparing direct and indirect evidence (Chen et al.). Vaginal misoprostol was found to be the treatment most likely to achieve vaginal delivery within 24 hours. Vaginal dinoprostone was found to be the next best. However, vaginal misoprostol was found to be most likely to be the worst treatment for increased hyperstimulation. Foley catheter was found to be the least likely to cause hyperstimulation, but is worse than both vaginal misoprostol and vaginal dinoprostone in achieving vaginal delivery within 24 hours. The two treatments that were most effective in achieving vaginal delivery within 24 hours have direct head-to-head trial data – and treatment effect obtained from the pairwise comparison is consistent with that obtained from the indirect comparison. These results are likely to be robust, at least for the main comparisons of interest. None declared. Completed disclosure of interests form available to view online as supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

PRM50 - A Frequentist Framework for Automated Generation of Node-Splitting Models for Assessment of Inconsistency in Network Meta-Analysis