Predicting hospital mortality among injured children using a national trauma database

Breaking barriers in trauma research: A narrative review of opportunities to leverage veterinary trauma for accelerated translation to clinical solutions for pets and people

Trauma is a common cause of morbidity and mortality in humans and companion animals. Recent efforts in procedural development, training, quality systems, data collection, and research have positively impacted patient outcomes; however, significant unmet need still exists. Coordinated efforts by collaborative, translational, multidisciplinary teams to advance trauma care and improve outcomes have the potential to benefit both human and veterinary patient populations. Strategic use of veterinary clinical trials informed by expertise along the research spectrum (i.e., benchtop discovery, applied science and engineering, large laboratory animal models, clinical veterinary studies, and human randomized trials) can lead to increased therapeutic options for animals while accelerating and enhancing translation by providing early data to reduce the cost and the risk of failed human clinical trials. Active topics of collaboration across the translational continuum include advancements in resuscitation (including austere environments), acute traumatic coagulopathy, trauma-induced coagulopathy, traumatic brain injury, systems biology, and trauma immunology. Mechanisms to improve funding and support innovative team science approaches to current problems in trauma care can accelerate needed, sustainable, and impactful progress in the field. This review article summarizes our current understanding of veterinary and human trauma, thereby identifying knowledge gaps and opportunities for collaborative, translational research to improve multispecies outcomes. This translational trauma group of MDs, PhDs, and DVMs posit that a common understanding of injury patterns and resulting cellular dysregulation in humans and companion animals has the potential to accelerate translation of research findings into clinical solutions.

Outcomes After Transfer of Pediatric Trauma Patients: Does Everyone Need to Visit the Trauma Bay?

INTRODUCTION

Critically injured children and teens often present to adult trauma centers or nontrauma facilities prior to transfer to a pediatric trauma center. For pediatric patients wanting transfer to the intensive care unit (ICU), there is little data to guide which can be safely transferred directly to the unit, and which should be evaluated first in the trauma bay.

METHODS

We used our institutional trauma registry to evaluate transferred trauma patients over a three year period. We compared time to imaging, time to operating room, and overall mortality between the group evaluated first in the emergency room and those transferred directly to the ICU.

RESULTS

When adjusted for other variables, there was no increased mortality in those transferred directly to the ICU. While there was a higher nonadjusted mortality in those transferred to the ICU (13% versus 3.7%), these nonsurvivors had a lower GCS (3 versus 13), higher Pediatric Risk of Mortality scores, and a high rate of severe head trauma. There was no significant delay in ordered imaging or procedures.

CONCLUSIONS

In patients, who have been assessed at another institution prior to transfer to the pediatric ICU, transfer directly to the ICU, bypassing the emergency department, does not delay interventions and does not appear to worsen outcomes.

Direct to operating room trauma resuscitation decreases mortality among severely injured children

BACKGROUND

Expediting evaluation and intervention for severely injured patients has remained a mainstay of advanced trauma care. One technique, direct to operating room (DOR) resuscitation, for selective adult patients has demonstrated decreased mortality. We sought to investigate the application of this protocol in children.

METHODS

All DOR pediatric patients from 2009 to 2016 at a pediatric Level I trauma center were identified. Direct to OR criteria included penetrating injury, chest injuries, amputations, significant blood loss, cardiopulmonary resuscitation, and surgeon discretion. Demographics, injury patterns, interventions, and outcomes were analyzed. Observed mortality was compared with expected mortality, calculated using Trauma Injury Severity Score methodology, with two-tailed t tests, and a p value less than 0.5 was considered significant.

RESULTS

Of 2,956 total pediatric trauma activations, 82 (2.8%) patients (age range, 1 month to 17 years) received DOR resuscitation during the study period. The most common indications for DOR were penetrating injuries (62%) and chest injuries (32%). Forty-four percent had Injury Severity Score (ISS) greater than 15, 33% had Glasgow Coma Scale (GCS) score of 8 or less, and 9% were hypotensive. The most commonly injured body regions were external (66%), head (34%), chest (30%), and abdomen (27%). Sixty-seven (82%) patients required emergent procedural intervention, most commonly wound exploration/repair (35%), central venous access (22%), tube thoracostomy (19%), and laparotomy (18%). Predictors of intervention were ISS greater than 15 (odds ratio, 14; p = 0.013) and GCS < 9 (odds ratio = 8.5, p = 0.044). The survival rate to discharge for DOR patients was 84% compared with an expected survival of 79% (Trauma Injury Severity Score) (p = 0.4). The greatest improvement relative to expected mortality was seen in the subgroup with penetrating trauma (84.5% vs 74.4%; p = 0.002).

CONCLUSION

A selective policy of resuscitating the most severely injured children in the OR can decrease mortality. Patients suffering penetrating trauma with the highest ISS, and diminished GCS scores have the greatest benefit. Trauma centers with appropriate resources should evaluate implementing similar policies.

LEVEL OF EVIDENCE

Diagnostic tests or criteria, level II.

Classifying, measuring and improving the quality of data in trauma registries: A review of the literature

INTRODUCTION

Globally, injury is a major cause of death and disability. Improvements in trauma care have been driven by trauma registries. The capacity of a trauma registry to inform improvements in the quality of trauma care is dependent upon the quality of data. The literature on data quality in disease registries is inconsistent and ambiguous; methods used for classifying, measuring, and improving data quality are not standardised. The aim of this study was to review the literature to determine the methods used to classify, measure and improve data quality in trauma registries.

METHODS

A scoping review of the literature was performed. Databases were searched using the term "trauma registry" and its synonyms, combined with multiple terms denoting data quality. There was no restriction on year. Full-length manuscripts were included if the classification, measurement or improvement of data quality in one or more trauma registries was a study objective. Data were abstracted regarding registry demographics, study design, data quality classification, and the reported methods used to measure and improve the pre-defined data quality dimensions of accuracy, completeness and capture.

RESULTS

Sixty-nine publications met the inclusion criteria. Four publications classified data quality. The most frequently described methods for measuring data accuracy (n=47) were checks against other datasets (n=18) and checks of injury coding (n=17). The most frequently described methods for measuring data completeness (n=47) were the percentage of included cases, for a given variable or list of variables, for which there was an observation in the registry (n=29). The most frequently described methods for measuring data capture (n=37) were the percentage of cases in a linked reference dataset that were also captured in the primary dataset being evaluated (n=24). Most publications dealing with the measurement of a dimension of data quality did not specify the methods used; most publications dealing with the improvement of data quality did not specify the dimension being targeted.

CONCLUSION

The classification, measurement and improvement of data quality in trauma registries is inconsistent. To maintain confidence in the usefulness of trauma registries, the metrics and reporting of data quality need to be standardised.

Use of Electronic Health-Related Datasets in Nursing and Health-Related Research

Datasets of gigabyte size are common in medical sciences. There is increasing consensus that significant untapped knowledge lies hidden in these large datasets. This review article aims to discuss Electronic Health-Related Datasets (EHRDs) in terms of types, features, advantages, limitations, and possible use in nursing and health-related research. Major scientific databases, MEDLINE, ScienceDirect, and Scopus, were searched for studies or review articles regarding using EHRDs in research. A total number of 442 articles were located. After application of study inclusion criteria, 113 articles were included in the final review. EHRDs were categorized into Electronic Administrative Health-Related Datasets and Electronic Clinical Health-Related Datasets. Subcategories of each major category were identified. EHRDs are invaluable assets for nursing the health-related research. Advanced research skills such as using analytical softwares, advanced statistical procedures, dealing with missing data and missing variables will maximize the efficient utilization of EHRDs in research.

Predicting in-hospital death among patients injured in traffic crashes in Saudi Arabia

INTRODUCTION

Traffic-related injuries are a major cause of premature death in developing countries. Saudi Arabia has struggled with high rates of traffic-related deaths for decades, yet little is known about health outcomes of motor vehicle victims seeking medical care. This study aims to develop and validate a model to predict in-hospital death among patients admitted to a large-urban trauma centre in Saudi Arabia for treatment following traffic-related crashes.

METHODS

The analysis used data from King Abdulaziz Medical City (KAMC) in Riyadh, Saudi Arabia. During the study period 2001-2010, 5325 patients met the inclusion criteria of being injured in traffic crashes and seen in the Emergency Department (ED) and/or admitted to the hospital. Backward stepwise logistic regression, with in-hospital death as the outcome, was performed. Variables with p<0.05 were included in the final model. The Bayesian Information Criterion (BIC) was employed to identify the most parsimonious model. Model discrimination was evaluated by the C-statistic and calibration by the Hosmer-Lemeshow Goodness of Fit statistic. Bootstrapping was used to assess overestimation of model performance and obtain a corrected C-statistic.

RESULTS

457 (8.5%) patients died at some time during their treatment in the ED or hospital. Older age, the Triage-Revised Trauma Scale (T-RTS), and Injury Severity Score were independent risk factors for in-hospital death: T-RTS was best modelled with linear and quadratic terms to capture a flattening of the relationship to death in the more severe range. The model showed excellent discrimination (C-statistic=0.96) and calibration (H-L statistic 4.29 [p>0.05]). Internal bootstrap validation gave similar results (C-statistic=0.96).

CONCLUSIONS

The proposed model can predict in-hospital death accurately. It can facilitate the triage process among injured patients, and identify unexpected deaths in order to address potential pitfalls in the care process. Conversely, by identifying high-risk patients, strategies can be developed to improve trauma care for these patients and reduce case-fatality. This is the first study to develop and validate a model to predict traffic-related mortality in a developing country. Future studies from developing countries can use this study as a reference for case fatality achievable for different risk profiles at a well-equipped trauma centre.

Helicopter overtriage in pediatric trauma

BACKGROUND

Helicopter Emergency Medical Services (HEMS) have been designed to provide faster access to trauma center care in cases of life-threatening injury. However, the ideal recipient population is not fully characterized, and indications for helicopter transport in pediatric trauma vary dramatically by county, state, and region. Overtriage, or unnecessary utilization, can lead to additional patient risk and expense. In this study we perform a nationwide descriptive analysis of HEMS for pediatric trauma and assess the incidence of overtriage in this group.

METHODS

We reviewed records from the American College of Surgeons National Trauma Data Bank (2008-11) and included patients less than 16 years of age who were transferred from the scene of injury to a trauma center via HEMS. Overtriage was defined as patients meeting all of the following criteria: Glasgow Coma Scale (GCS) equal to 15, absence of hypotension, an Injury Severity Score (ISS) less than 9, no need for procedure or critical care, and a hospital length of stay of less than 24 hours.

RESULTS

A total of 19,725 patients were identified with a mean age of 10.5 years. The majority of injuries were blunt (95.6%) and resulted from motor vehicle crashes (48%) and falls (15%). HEMS transported patients were predominately normotensive (96%), had a GCS of 15 (67%), and presented with minor injuries (ISS<9, 41%). Overall, 28 % of patients stayed in the hospital for less than 24 hours, and the incidence of overtriage was 17%.

CONCLUSIONS

Helicopter overtriage is prevalent among pediatric trauma patients nationwide. The ideal model to predict need for HEMS must consider clinical outcomes in the context of judicious resource utilization. The development of guidelines for HEMS use in pediatric trauma could potentially limit unnecessary transfers while still identifying children who require trauma center care in a timely fashion.

Comparison of outcomes for children with cervical spine injury based on destination hospital from scene of injury

BACKGROUND

Pediatric cervical spine injury is rare. As a result, evidence-based guidance for prehospital triage of children with suspected cervical spine injuries is limited. The effects of transport time and secondary transfer for specialty care have not previously been examined in the subset of children with cervical spine injuries.

OBJECTIVES

The primary objective was to determine if prehospital destination choice affects outcomes for children with cervical spine injuries. The secondary objectives were to describe prehospital and local hospital interventions for children ultimately transferred to pediatric trauma centers for definitive care of cervical spine injuries.

METHODS

The authors searched the Pediatric Emergency Care Applied Research Network (PECARN) cervical spine injury data set for children transported by emergency medical services (EMS) from scene of injury. Neurologic outcomes in children with cervical spine injuries transported directly to pediatric trauma centers were compared with those transported to local hospitals and later transferred to pediatric trauma centers, adjusting for injury severity, indicated by altered mental status, focal neurologic deficits, and substantial comorbid injuries. In addition, transport times and interventions provided in the prehospital, local hospital, and pediatric trauma center settings were compared. Multiple imputation was used to handle missing data.

RESULTS

The PECARN cervical spine injury cohort contains 364 patients transported from scene of injury by EMS. A total of 321 met our inclusion criteria. Of these, 180 were transported directly to pediatric trauma centers, and 141 were transported to local hospitals and later transferred. After adjustments for injury severity, odds of a normal outcome versus death or persistent neurologic deficit were higher for patients transported directly to pediatric trauma centers (odds ratio [OR] = 1.89, 95% confidence interval [CI] = 1.03 to 3.47). EMS transport times to first hospital did not differ and did not affect outcomes. Prehospital analgesia was very infrequent.

CONCLUSIONS

Initial destination from scene (pediatric trauma center vs. local hospital) appears to be associated with neurologic outcome of children with cervical spine injuries. Markers of injury severity (altered mental status and focal neurologic findings) are important predictors of poor outcome in children with cervical spine injuries and should remain the primary guide for prehospital triage to designated trauma centers.

An Overview of Administrative and National Survey Databases for Use in Otolaryngology Research

An emerging focus on measurement and reporting of health care quality indicators calls for a sound evidence base that otolaryngologists can use to support clinical care decisions. In addition to traditional clinical trials, practitioners are heeding results of high-quality comparative-effectiveness and health services research analyses to better understand the complexity of disease epidemiology, care variation, health care costs, and surgical utilization for common conditions. Many national publicly available data sources exist for use in clinical research that may be of benefit for the academic and clinical otolaryngologist. The strength and value of these sources vary depending on the intended use or research question. The purpose of this commentary is to introduce and provide an overview of some major national and administrative databases, highlight potential strengths and limitations of these data, and suggest applications for use in advancing the care of our patients in otolaryngology.

Blast injury in children: an analysis from Afghanistan and Iraq, 2002-2010

BACKGROUND

Throughout history, children have been victims of armed conflict, including the blast injury complex, however, the pattern of injury, physiologic impact, and treatment needs of children with this injury are not well documented.

METHODS

The Joint Theatre Trauma Registry provides data on all civilians admitted to US military treatment facilities from 2002 to 2010 with injuries from an explosive device. The data were stratified by age and analyzed for differences in anatomic injury patterns, Injury Severity Score (ISS), Revised Trauma Score (RTS), mortality, intensive care unit days, and length of hospitalization. Multivariate logistic regression was done to determine independent predictors of mortality. All operative procedures with a specified site were tabulated and categorized by body region and age.

RESULTS

A total of 4,983 civilian patients were admitted, 25% of whom were younger than 15 years. Pediatric patients aged 8 to 14 years had a higher ISS and hospital stay than other age groups, and children younger than 15 years had a longer intensive care unit stay. Injuries in children were more likely to occur in the head and neck and less likely in the bony pelvis and extremities. Children had a lower RTS than the other age groups. Mortality correlated highly with burns, head injury, transfusion, and RTS. Adolescent patients had a lower mortality rate than the other age groups. Improvised explosive devices were the most common cause of injury in all age groups.

CONCLUSION

Children experiencing blast injury complex have an anatomic pattern that is unique and an RTS that reflects more severe physiologic derangement. Injuries requiring transfusion or involving the head and neck and burns were predictive of mortality, and this persisted across all age groups. The mortality rate of children with blast injury is significant (7%), and treatment is resource intensive, requiring many surgical subspecialties.

LEVEL OF EVIDENCE

Epidemiologic study, level III.

Analysis of clinical and ultrasonographic data by use of logistic regression models for prediction of malignant versus benign causes of ultrasonographically detected focal liver lesions in dogs

OBJECTIVE

To investigate the value of clinical, laboratory, and imaging data for use in predicting malignant or benign histologic results for ultrasonographically detected focal liver lesions in dogs.

SAMPLE

Records and archived images of 247 dogs evaluated at the University of Minnesota Veterinary Medical Center from 2005 to 2008 that underwent abdominal ultrasonography and histologic evaluation of the liver.

PROCEDURES

Data were analyzed with multivariable logistic regression models. All dogs were classified as having benign or malignant liver disease on the basis of histologic reports. Three multivariable logistic regression models were fit to a development subset of the data by use of combinations of signalment, historical, physical examination, laboratory, and diagnostic imaging (survey radiography and abdominal ultrasonography) data as predictor variables. The resulting models were validated by evaluating predictive performance against a holdout validation subset of the data.

RESULTS

Models that included ultrasonographic variables had the highest overall predictive value. In these models, greater lesion size and the presence of peritoneal fluid were the only variables that had a positive association with malignant liver disease.

CONCLUSIONS AND CLINICAL RELEVANCE

Large ultrasonographically detected liver lesions and the presence of peritoneal fluid were associated with malignant liver disease in dogs.

Influence of the National Trauma Data Bank on the study of trauma outcomes: is it time to set research best practices to further enhance its impact?

BACKGROUND

Risk-adjusted analyses are critical in evaluating trauma outcomes. The National Trauma Data Bank (NTDB) is a statistically robust registry that allows such analyses; however, analytical techniques are not yet standardized. In this study, we examined peer-reviewed manuscripts published using NTDB data, with particular attention to characteristics strongly associated with trauma outcomes. Our objective was to determine if there are substantial variations in the methodology and quality of risk-adjusted analyses and therefore, whether development of best practices for risk-adjusted analyses is warranted.

STUDY DESIGN

A database of all studies using NTDB data published through December 2010 was created by searching PubMed and Embase. Studies with multivariate risk-adjusted analyses were examined for their central question, main outcomes measures, analytical techniques, covariates in adjusted analyses, and handling of missing data.

RESULTS

Of 286 NTDB publications, 122 performed a multivariable adjusted analysis. These studies focused on clinical outcomes (51 studies), public health policy or injury prevention (30), quality (16), disparities (15), trauma center designation (6), or scoring systems (4). Mortality was the main outcome in 98 of these studies. There were considerable differences in the covariates used for case adjustment. The 3 covariates most frequently controlled for were age (95%), Injury Severity Score (85%), and sex (78%). Up to 43% of studies did not control for the 5 basic covariates necessary to conduct a risk-adjusted analysis of trauma mortality. Less than 10% of studies used clustering to adjust for facility differences or imputation to handle missing data.

CONCLUSIONS

There is significant variability in how risk-adjusted analyses using data from the NTDB are performed. Best practices are needed to further improve the quality of research from the NTDB.

Pediatric trauma mortality by type of designated hospital in a mature inclusive trauma system

Background:

Previous studies have shown divergent results regarding the survival of injured children treated at pediatric trauma centers (PTC) and adult trauma centers (ATC).

Aims:

(1) To document, in a regionalized inclusive trauma system, at which level of trauma centers were the injured children treated and (2) to compare the in-hospital mortality over five levels of trauma care, ranging from pediatric level I trauma centers (PTC) to designated local trauma hospitals (level IV) for the whole study sample and for subgroups of severely injured children and head trauma.

Materials and Methods:

A retrospective analysis included data on 11,053 injured children (age ≤16 years) treated between April 1998 and March 2005 in 58 designated trauma hospitals in the province of Quebec, Canada. Multiple imputation was used to handle missing physiological data and multivariate logistic regression was used to compare mortality over levels of care.

Results:

PTC treated 52.2% of the children. Children treated at PTC were more often transferred from another hospital (73%) and were more severely injured. ATC level I, II, III and IV centers treated, respectively, 3.0%, 16.2%, 24.3% and 4.3% of children. Compared with children treated at a PTC, the risk of mortality was higher for children treated at each other ATC, i.e. level I (adjusted odds ratio [OR] = 3.1; 95% confidence interval [CI]: 1.3–7.5), level II (OR = 2.5; 95% CI: 1.3–5.0), level III (OR = 5.2; 95% CI: 2.1–13.1) and level IV (OR = 9.9; 95% CI: 2.4–41.3). Similar findings were observed among the subsamples of children who were more severely injured (Injury Severity Score >15) and who sustained head injuries.

Conclusions:

In our trauma system, PTC cared for more than half of the injured children and patients treated there have better survival than those treated at all other levels of ATC.

Visualizing and assessing discrimination in the logistic regression model

Logistic regression models are widely used in medicine for predicting patient outcome (prognosis) and constructing diagnostic tests (diagnosis). Multivariable logistic models yield an (approximately) continuous risk score, a transformation of which gives the estimated event probability for an individual. A key aspect of model performance is discrimination, that is, the model's ability to distinguish between patients who have (or will have) an event of interest and those who do not (or will not). Graphical aids are important in understanding a logistic model. The receiver-operating characteristic (ROC) curve is familiar, but not necessarily easy to interpret. We advocate a simple graphic that provides further insight into discrimination, namely a histogram or dot plot of the risk score in the outcome groups. The most popular performance measure for the logistic model is the c-index, numerically equivalent to the area under the ROC curve. We discuss the comparative merits of the c-index and the (standardized) mean difference in risk score between the outcome groups. The latter statistic, sometimes known generically as the effect size, has been computed in slightly different ways by several different authors, including Glass, Cohen and Hedges. An alternative measure is the overlap between the distributions in the outcome groups, defined as the area under the minimum of the two density functions. The larger the overlap, the weaker the discrimination. Under certain assumptions about the distribution of the risk score, the c-index, effect size and overlap are functionally related. We illustrate the ideas with simulated and real data sets.

The availability and use of out-of-hospital physiologic information to identify high-risk injured children in a multisite, population-based cohort

OBJECTIVE

The validity of using adult physiologic criteria to triage injured children in the out-of-hospital setting remains unproven. Among children meeting adult field physiologic criteria, we assessed the availability of physiologic information, the incidence of death or prolonged hospitalization, and whether age-specific criteria would improve the specificity of the physiologic triage step.

METHODS

We analyzed a prospective, out-of-hospital cohort of injured children aged < or =14 years collected from December 2005 through February 2007 by 237 emergency medical services (EMS) agencies transporting to 207 acute care hospitals (trauma and nontrauma centers) in 11 sites across the United States and Canada. Inclusion criteria were standard adult physiologic values: systolic blood pressure (SBP) < or =90 mmHg, respiratory rate < 10 or > 29 breaths/min, Glasgow Coma Scale (GCS) score < or =12, and field intubation attempt. Seven physiologic variables (including age-specific values) and three demographic and mechanism variables were included in the analysis. "High-risk" children included those who died (field or in-hospital) or were hospitalized > 2 days. The decision tree was derived and validated using binary recursive partitioning.

RESULTS

Nine hundred fifty-five children were included in the analysis, of whom 62 (6.5%) died and 117 (12.3%) were hospitalized > 2 days. Missing values were common, ranging from 6% (respiratory rate) to 53% (pulse oximetry), and were associated with younger age and high-risk outcome. The final decision rule included four variables (assisted ventilation, GCS score < 11, pulse oximetry < 95%, and SBP > 96 mmHg), which demonstrated improved specificity (71.7% [95% confidence interval (CI) 66.7-76.6%]) at the expense of missing high-risk children (sensitivity 76.5% [95% CI 66.4-86.6%]).

CONCLUSIONS

The incidence of high-risk injured children meeting adult physiologic criteria is relatively low and the findings from this sample do not support using age-specific values to better identify such children. However, the amount and pattern of missing data may compromise the value and practical use of field physiologic information in pediatric triage.

A multiple imputation model for imputing missing physiologic data in the national trauma data bank

BACKGROUND

Like most trauma registries, the National Trauma Data Bank (NTDB) is limited by the problem of missing physiologic data. Multiple imputation (MI) has been proposed to simulate missing Glasgow Coma Scale (GCS) scores, respiratory rate (RR), and systolic blood pressure (SBP). The aim of this study was to develop an MI model for missing physiologic data in the NTDB and to provide guidelines for its implementation.

STUDY DESIGN

The NTDB 7.0 was restricted to patients admitted in 2005 with at least one anatomic injury code. A series of auxiliary variables thought to offer information for the imputation process was selected from the NTDB by literature review and expert opinion. The relation of these variables to physiologic variables and to the fact that they were missing was examined using logistic regression. The MI model included all auxiliary variables that had a statistically significant association with physiologic variables or with the fact that they were missing (Bonferroni-corrected p value <0.05).

RESULTS

The NTDB sample included 373,243 observations. Glasgow Coma Scale, respiratory rate, and systolic blood pressure were missing for 20.3%, 3.9%, and 8.5% of data observations, respectively. The MI model included information on the following: gender, age, anatomic injury severity, transfer status, injury mechanism, intubation status, alcohol and drug test results, emergency department disposition, total length of stay, ICU length of stay, duration of mechanical ventilation, and discharge disposition. The MI model offered good discrimination for predicting the value of physiologic variables and the fact that they were missing (areas under the receiver operating characteristic curve between 0.832 and 0.999).

CONCLUSIONS

This article proposes an MI model for imputing missing physiologic data in the NTDB and provides guidelines to facilitate its use. Implementation of the model should improve the quality of research involving the NTDB. The methodology can also be adapted to other trauma registries.

Development and validation of a prediction model with missing predictor data: a practical approach

OBJECTIVE

To illustrate the sequence of steps needed to develop and validate a clinical prediction model, when missing predictor values have been multiply imputed.

STUDY DESIGN AND SETTING

We used data from consecutive primary care patients suspected of deep venous thrombosis (DVT) to develop and validate a diagnostic model for the presence of DVT. Missing values were imputed 10 times with the MICE conditional imputation method. After the selection of predictors and transformations for continuous predictors according to three different methods, we estimated regression coefficients and performance measures.

RESULTS

The three methods to select predictors and transformations of continuous predictors showed similar results. Rubin's rules could easily be applied to estimate regression coefficients and performance measures, once predictors and transformations were selected.

CONCLUSION

We provide a practical approach for model development and validation with multiply imputed data.

Despite overall low pediatric head injury mortality, disparities exist between races

BACKGROUND

To continually improve quality of care, it is important for centers caring for children with head injury to evaluate their overall outcome and that among diverse patient groups.

METHODS

Data on children with head injuries were extracted from the National Trauma Data Bank of the American College of Surgeons and our local trauma registry. Unadjusted mortality, as well as stratified analysis and logistic regression modeling, was used to evaluate overall and race-specific mortality.

RESULTS

There were 13,363 children in the National Trauma Data Base and 3111 in our database included. Our overall mortality rate compared favorably with the national data (3.2% vs 6.8%, P < .05). Our local data, however, showed a significant difference in mortality between white and African American (AA) children (2.2% vs 5.3%, P < .05), which was not identified in the national data. After stratification, the disparities by race persisted. Finally, multivariate regression modeling revealed that AA race was an independent predictor of mortality among our patient population, with an odds ratio of 3.1 (95% confidence interval, 1.2-7.8).

CONCLUSION

Despite excellent outcomes for children with head injuries, we have uncovered unsettling inequities between AA and white children. These findings support the need to evaluate outcomes among specific groups to identify disparities that require further careful investigation.

The Resuscitation Outcomes Consortium Epistry-Trauma: design, development, and implementation of a North American epidemiologic prehospital trauma registry

Injury is a major public health problem generating substantial morbidity, mortality, and economic burden on society. The majority of seriously injured persons are initially evaluated and cared for by prehospital providers, however the effect of emergency medical services (EMS) systems, EMS clinical care, and EMS interventions on trauma patient outcomes is largely unknown. Outcome-based information to guide future EMS care has been hampered by the lack of comprehensive, standardized, multi-center prehospital data resources that include meaningful patient outcomes. In this paper, we describe the background, design, development, implementation, content, and potential uses of the first North American comprehensive epidemiologic prehospital data registry for injured persons. This data registry samples patients from 264 EMS agencies transporting to 287 acute care hospitals in both the United States and Canada.

Bayesian logistic injury severity score: a method for predicting mortality using international classification of disease-9 codes

OBJECTIVES

Owing to the large number of injury International Classification of Disease-9 revision (ICD-9) codes, it is not feasible to use standard regression methods to estimate the independent risk of death for each injury code. Bayesian logistic regression is a method that can select among a large numbers of predictors without loss of model performance. The purpose of this study was to develop a model for predicting in-hospital trauma deaths based on this method and to compare its performance with the ICD-9-based Injury Severity Score (ICISS).

METHODS

The authors used Bayesian logistic regression to train and test models for predicting mortality based on injury ICD-9 codes (2,210 codes) and injury codes with two-way interactions (243,037 codes and interactions) using data from the National Trauma Data Bank (NTDB). They evaluated discrimination using area under the receiver operating curve (AUC) and calibration with the Hosmer-Lemeshow (HL) h-statistic. The authors compared performance of these models with one developed using ICISS.

RESULTS

The discrimination of a model developed using individual ICD-9 codes was similar to that of a model developed using individual codes and their interactions (AUC = 0.888 vs. 0.892). Inclusion of injury interactions, however, improved model calibration (HL h-statistic = 2,737 vs. 1,347). A model based on ICISS had similar discrimination (AUC = .855) but showed worse calibration (HL h-statistic = 45,237) than those based on regression.

CONCLUSIONS

A model that incorporates injury interactions had better predictive performance than one based only on individual injuries. A regression approach to predicting injury mortality based on injury ICD-9 codes yields models with better predictive performance than ICISS.

Using multiple imputation and propensity scores to test the effect of car seats and seat belt usage on injury severity from trauma registry data

BACKGROUND

Missing data and the retrospective, nonrandomized nature of trauma registries can decrease the quality of registry-based research. Therefore, we used multiple imputation and propensity scores to test the effect of car seats and seat belt usage on injury severity in children involved in motor vehicle crashes.

METHODS

All children admitted after injury from motor vehicle crashes who had complete data on seat belt or car seat usage from 2003 to 2006 were included in the study. The sample was divided into children younger than 4 years (n = 130) or 5 years or older (n = 575) and analyzed for seat belt usage, car seat usage, injury severity score, revised trauma score, and Glasgow Coma Scale score. Data were analyzed before and after matching on propensity scores after multiple imputation.

RESULTS

There were no outcome differences between car seat users and non-car seat users. However, there were significant improvements in injury severity score (7.0 vs. 10.1, P = .002) and revised trauma score (7.6 vs 7.3, P = .013 for seat belt users compared to nonusers) even after matching on propensity score.

CONCLUSION

Multiple imputation and propensity scores demonstrated the efficacy of seat belts, but not car seat in this preliminary study. This statistical method can strengthen registry-based research.

The predictive value and appropriate ranges of prehospital physiological parameters for high-risk injured children

OBJECTIVES

To assess: (1) the relative importance of prehospital physiological measures in identifying high-risk children; (2) whether different age-based criteria should be used for each prehospital physiological measure; and (3) outcome-based appropriate ranges of physiological measures in injured children.

METHODS

This was a retrospective cohort analysis of injured children 0 to 14 years transported by emergency medical services to 48 statewide hospitals from January 1, 1998, through December 31, 2003. We analyzed prehospital physiological measures, including Glasgow Coma Scale score (GCS), systolic blood pressure (SBP), respiratory rate (RR), heart rate, shock index (heart rate/SBP), and airway intervention. "High-risk" children were defined as those with in-hospital mortality, major nonorthopedic surgery, intensive care unit stay greater than or equal to 2 days, or Injury Severity Score greater than or equal to 16. Specific age groups included 0 to 2 years, 3 to 5 years, 6 to 10 years, and 11 to 14 years.

RESULTS

A total of 3877 injured children were included in the analysis, of which 1111 (29%) were high risk. Prehospital GCS was the variable of greatest importance in identifying high-risk children, followed by (in order) airway intervention, RR, heart rate, SBP, and shock index. Age modified the effect of prehospital RR (P = 0.0046), heart rate (P = 0.01), and SBP (P = 0.02). There was a linear relationship between GCS and outcome that was consistent across all ages. Specific age-based ranges of other physiological measures were identified for high-risk children.

CONCLUSIONS

Prehospital GCS and respiratory compromise were the most important physiological measures in identifying high-risk injured children. Age-specific criteria should be considered for RR, heart rate, and SBP.

Disparities in child abuse mortality are not explained by injury severity

BACKGROUND

Unadjusted abuse-related mortality has been demonstrated to be nearly 4-fold higher for African American (AA) children. Little is known about the etiology of this disparity. This study examines the importance of injury severity and initial presentation in explaining the observed disparity.

METHOD

Our trauma database was reviewed to identify all abused patients admitted over a 10-year period. Outcomes among white and AA children were compared with specific attention to injury severity scores and initial presentation. Logistic regression and Cox proportional hazard analyses were performed to evaluate the impact of race on outcome.

RESULTS

There were 443 abused children identified. Thirty-eight percent of the group was AA. The overall mortality was 7.7%; however, the AA mortality was significantly higher than white children (14.8% vs 3.3%; P < .05). After controlling for injury severity and physiology at presentation, the odds ratio of mortality for an AA child was 9.14 (95% confidence interval, 1.97-42.43). Survival analysis confirmed the disparity after revealing a hazard ratio of dying for AA children of 6.51 (95% confidence interval, 2.74-15.47) compared with white children.

CONCLUSION

Despite attempts to control for the clinical presentation and injury severity of abused children, significant differences in mortality persist between AA and white children.