Classification and regression trees (CART) for prediction of function at 1 year following head trauma

Data mining and mathematical models in cancer prognosis and prediction

Cancer is a fetal and complex disease. Individual differences of the same cancer type or the same patient at different stages of cancer development may require distinct treatments. Pathological differences are reflected in tissues, cells and gene levels etc. The interactions between the cancer cells and nearby microenvironments can also influence the cancer progression and metastasis. It is a huge challenge to understand all of these mechanistically and quantitatively. Researchers applied pattern recognition algorithms such as machine learning or data mining to predict cancer types or classifications. With the rapidly growing and available computing powers, researchers begin to integrate huge data sets, multi-dimensional data types and information. The cells are controlled by the gene expressions determined by the promoter sequences and transcription regulators. For example, the changes in the gene expression through these underlying mechanisms can modify cell progressing in the cell-cycle. Such molecular activities can be governed by the gene regulations through the underlying gene regulatory networks, which are essential for cancer study when the information and gene regulations are clear and available. In this review, we briefly introduce several machine learning methods of cancer prediction and classification which include Artificial Neural Networks (ANNs), Decision Trees (DTs), Support Vector Machine (SVM) and naive Bayes. Then we describe a few typical models for building up gene regulatory networks such as Correlation, Regression and Bayes methods based on available data. These methods can help on cancer diagnosis such as susceptibility, recurrence, survival etc. At last, we summarize and compare the modeling methods to analyze the development and progression of cancer through gene regulatory networks. These models can provide possible physical strategies to analyze cancer progression in a systematic and quantitative way.

Hypertension-Related Drug Activity Identification Based on Novel Ensemble Method

Hypertension is a chronic disease and major risk factor for cardiovascular and cerebrovascular diseases that often leads to damage to target organs. The prevention and treatment of hypertension is crucially important for human health. In this paper, a novel ensemble method based on a flexible neural tree (FNT) is proposed to identify hypertension-related active compounds. In the ensemble method, the base classifiers are Multi-Grained Cascade Forest (gcForest), support vector machines (SVM), random forest (RF), AdaBoost, decision tree (DT), Gradient Boosting Decision Tree (GBDT), KNN, logical regression, and naïve Bayes (NB). The classification results of nine classifiers are utilized as the input vector of FNT, which is utilized as a nonlinear ensemble method to identify hypertension-related drug compounds. The experiment data are extracted from hypertension-unrelated and hypertension-related compounds collected from the up-to-date literature. The results reveal that our proposed ensemble method performs better than other single classifiers in terms of ROC curve, AUC, TPR, FRP, Precision, Specificity, and F1. Our proposed method is also compared with the averaged and voting ensemble methods. The results reveal that our method could identify hypertension-related compounds more accurately than two classical ensemble methods.

An Interpretable Machine Learning Model to Classify Coronary Bifurcation Lesions

Coronary bifurcation lesions are a leading cause of Coronary Artery Disease (CAD). Despite its prevalence, coronary bifurcation lesions remain difficult to treat due to our incomplete understanding of how various features of lesion anatomy synergistically disrupt normal hemodynamic flow. In this work, we employ an interpretable machine learning algorithm, the Classification and Regression Tree (CART), to model the impact of these geometric features on local hemodynamic quantities. We generate a synthetic arterial database via computational fluid dynamic simulations and apply the CART approach to predict the time averaged wall shear stress (TAWSS) at two different locations within the cardiac vasculature. Our experimental results show that CART can estimate a simple, interpretable, yet accurately predictive nonlinear model of TAWSS as a function of such features.Clinical relevance- The fitted tree models have the potential to refine predictions of disturbed hemodynamic flow based on an individual's cardiac and lesion anatomy and consequently makes progress towards personalized treatment planning for CAD patients.

Using Decision Tree Methodology to Predict Employment After Moderate to Severe Traumatic Brain Injury

Objective:

To build decision tree prediction models for long-term employment outcomes of individuals after moderate to severe closed traumatic brain injury (TBI) and assess model accuracy in an independent sample.

Setting:

TBI Model Systems Centers.

Participants:

TBI Model Systems National Database participants injured between January 1997 and January 2017 with moderate to severe closed TBI. Sample sizes were 7867 (year 1 postinjury), 6783 (year 2 postinjury), and 4927 (year 5 postinjury).

Design:

Cross-sectional analyses using flexible classification tree methodology and validation using an independent subset of TBI Model Systems National Database participants.

Main Measures:

Competitive employment at 1, 2, and 5 years postinjury.

Results:

In the final employment prediction models, posttraumatic amnesia duration was the most important predictor of employment in each outcome year. Additional variables consistently contributing were age, preinjury education, productivity, and occupational category. Generally, individuals spending fewer days in posttraumatic amnesia, who were competitively employed preinjury, and more highly educated had better outcomes. Predictability in test data sets ranged from a C-statistic of 0.72 (year 5; confidence interval: 0.68-0.76) to 0.77 (year 1; confidence interval: 0.74-0.80).

Conclusion:

An easy-to-use decision tree tool was created to provide prognostic information on long-term competitive employment outcomes in individuals with moderate to severe closed TBI. Length of posttraumatic amnesia, a clinical marker of injury severity, and preinjury education and employment status were the most important predictors.

Comparison of Different Cropland Classification Methods under Diversified Agroecological Conditions in the Zambezi River Basin

Having updated knowledge of cropland extent is essential for crop monitoring and food security early warning. Previous research has proposed different methods and adopted various datasets for mapping cropland areas at regional to global scales. However, most approaches did not consider the characteristics of farming systems and apply the same classification method in different agroecological zones (AEZs). Furthermore, the acquisition of in situ samples for classification training remains challenging. To address these knowledge gaps and challenges, this study applied a zone-specific classification by comparing four classifiers (random forest, the support vector machine (SVM), the classification and regression tree (CART) and minimum distance) for cropland mapping over four different AEZs in the Zambezi River basin (ZRB). Landsat-8 and Sentinel-2 data and derived indices were used and synthesized to generate thirty-five layers for classification on the Google Earth Engine platform. Training samples were derived from three existing landcover datasets to minimize the cost of sample acquisitions over the large area. The final cropland map was generated at a 10 m resolution. The performance of the four classifiers and the viability of training samples were analysed. All classifiers presented higher accuracy in cool AEZs than in warm AEZs, which may be attributed to field size and lower confusion between cropland and grassland classes. This indicates that agricultural landscape may impact classification results regardless of the classifiers. Random forest was found to be the most stable and accurate classifier across different agricultural systems, with an overall accuracy of 84% and a kappa coefficient of 0.67. Samples extracted over the full agreement areas among existing datasets reduced uncertainty and provided reliable calibration sets as a replacement of costly in situ measurements. The methodology proposed by this study can be used to generate periodical high-resolution cropland maps in ZRB, which is helpful for the analysis of cropland extension and abandonment as well as intensity changes in response to the escalating population and food insecurity.

Earth fissure hazard prediction using machine learning models

Earth fissures are the cracks on the surface of the earth mainly formed in the arid and the semi-arid basins. The excessive withdrawal of groundwater, as well as the other underground natural resources, has been introduced as the significant causing of land subsidence and potentially, the earth fissuring. Fissuring is rapidly turning into the nations' major disasters which are responsible for significant economic, social, and environmental damages with devastating consequences. Modeling the earth fissure hazard is particularly important for identifying the vulnerable groundwater areas for the informed water management, and effectively enforce the groundwater recharge policies toward the sustainable conservation plans to preserve existing groundwater resources. Modeling the formation of earth fissures and ultimately prediction of the hazardous areas has been greatly challenged due to the complexity, and the multidisciplinary involved to predict the earth fissures. This paper aims at proposing novel machine learning models for prediction of earth fissuring hazards. The Simulated annealing feature selection (SAFS) method was applied to identify key features, and the generalized linear model (GLM), multivariate adaptive regression splines (MARS), classification and regression tree (CART), random forest (RF), and support vector machine (SVM) have been used for the first time to build the prediction models. Results indicated that all the models had good accuracy (>86%) and precision (>81%) in the prediction of the earth fissure hazard. The GLM model (as a linear model) had the lowest performance, while the RF model was the best model in the modeling process. Sensitivity analysis indicated that the hazardous class in the study area was mainly related to low elevations with characteristics of high groundwater withdrawal, drop in groundwater level, high well density, high road density, low precipitation, and Quaternary sediments distribution.

Patients with minor diseases who access high-tier medical care facilities: New evidence from classification and regression trees

OBJECTIVES

Patients in Taiwan's National Health Insurance (NHI) program can choose a medical care facility of any tier for outpatient visits, without a referral. However, this system results in high medical expenditures and costs of outpatient visits. In this study, patients who had only minor diseases but who accessed high-tier medical care facilities were investigated using classification and regression trees.

METHODS

For this study, data were obtained from the Taiwan NHI Research Database. First, 280 diseases, coded according to the Clinical Classification Software (CCS), were examined to determine whether patients chose the most appropriate facility when seeking medical care. After controlling for the CCS codes, an investigation into the types of patients who visit high-tier medical care facilities was conducted.

RESULTS

Chronic disease status and CCS code were critical for constructing the classification trees. Male patients living in urban areas and earning a higher income were more likely to access high-tier medical care facilities. However, changes to the NHI copayment policies have significantly reduced the probability of utilizing high-tier medical care facilities.

CONCLUSIONS

Factors relevant to patients' selection of high-tier medical care facilities were identified. Overall, increasing patients' out-of-pocket payments significantly reduced the probability of accessing high-tier medical facilities.

A Detailed Overview of Long-Term Outcomes in Severe Traumatic Brain Injury Eight Years Post-injury

Background and aims: Severe traumatic brain injury is a leading cause of acquired persistent disabilities, and represents an important health and economic burden. However, the determinants of long-term outcome have rarely been systematically studied in a prospective longitudinal study of a homogeneous group of patients suffering exclusively from severe TBI

Methods: Prospective observational study of an inception cohort of adult patients with severe traumatic brain injury in the Parisian area (PariS-TBI). Outcome was assessed with face-to-face interview 8 years after Traumatic Brain Injury, focusing on impairments, activity limitations, and participation restriction.

Results: Five hundred and four patients were included between 2005 and 2007. At 8-year follow-up, 261 patients were deceased, 128 were lost to follow-up, 22 refused to participate, and 86 were finally evaluated. Age, gender, initial injury severity did not significantly differ between evaluated patients and lost to follow-up, but the latter were more frequently students or unemployed. Mean age was 41.9 (SD 13.6), 79% were male, median initial Glasgow Coma Scale Score was 6. The most frequent somatic complaints concerned balance (47.5%), motricity (31%), and headaches (36%), but these were less frequent than cognitive complaints (Memory 71%, Slowness 68%, Concentration 67%). According to the Hospital Anxiety and Depression Scale (HADS), 25 % had a score >8 for anxiety and 23.7% for depression. According to the Extended Glasgow Outcome Scale, 19.8% remained severely disabled, 46.5% moderately disabled, 33.7% had a good recovery. Older age, longer education duration, lower functional status upon intensive care discharge, and more severe 8-year dysexecutive problems were significantly associated with a lower Extended Glasgow Outcome Scale score in multivariable analysis. At 8 years, 48.7% of patients were employed in a productive job. Of those, 38% declared a salary loss since traumatic brain injury. Unemployment was significantly associated with lower 1-year GOSE score and more severe 8-year dysexecutive problems.

Conclusions: These results from an inception cohort study highlight the fact that long-term outcome after severe TBI is determined by a complex combination of injury-related, demographic and neuropsychological factors. Long after the injury, persisting impairments still interfere with social integration, and participation.

Simplifying the use of prognostic information in traumatic brain injury. Part 2: Graphical presentation of probabilities

OBJECTIVE Clinical features such as those included in the Glasgow Coma Scale (GCS) score, pupil reactivity, and patient age, as well as CT findings, have clear established relationships with patient outcomes due to neurotrauma. Nevertheless, predictions made from combining these features in probabilistic models have not found a role in clinical practice. In this study, the authors aimed to develop a method of displaying probabilities graphically that would be simple and easy to use, thus improving the usefulness of prognostic information in neurotrauma. This work builds on a companion paper describing the GCS-Pupils score (GCS-P) as a tool for assessing the clinical severity of neurotrauma. METHODS Information about early GCS score, pupil response, patient age, CT findings, late outcome according to the Glasgow Outcome Scale, and mortality were obtained at the individual adult patient level from the CRASH (Corticosteroid Randomisation After Significant Head Injury; n = 9045) and IMPACT (International Mission for Prognosis and Clinical Trials in TBI; n = 6855) databases. These data were combined into a pooled data set for the main analysis. Logistic regression was first used to model the combined association between the GCS-P and patient age and outcome, following which CT findings were added to the models. The proportion of variability in outcomes "explained" by each model was assessed using Nagelkerke's R². RESULTS The authors observed that patient age and GCS-P have an additive effect on outcome. The probability of mortality 6 months after neurotrauma is greater with increasing age, and for all age groups the probability of death is greater with decreasing GCS-P. Conversely, the probability of favorable recovery becomes lower with increasing age and lessens with decreasing GCS-P. The effect of combining the GCS-P with patient age was substantially more informative than the GCS-P, age, GCS score, or pupil reactivity alone. Two-dimensional charts were produced displaying outcome probabilities, as percentages, for 5-year increments in age between 15 and 85 years, and for GCS-Ps ranging from 1 to 15; it is readily seen that the movement toward combinations at the top right of the charts reflects a decreasing likelihood of mortality and an increasing likelihood of favorable outcome. Analysis of CT findings showed that differences in outcome are very similar between patients with or without a hematoma, absent cisterns, or subarachnoid hemorrhage. Taken in combination, there is a gradation in risk that aligns with increasing numbers of any of these abnormalities. This information provides added value over age and GCS-P alone, supporting a simple extension of the earlier prognostic charts by stratifying the original charts in the following 3 CT groupings: none, only 1, and 2 or more CT abnormalities. CONCLUSIONS The important prognostic features in neurotrauma can be brought together to display graphically their combined effects on risks of death or on prospects for independent recovery. This approach can support decision making and improve communication of risk among health care professionals, patients, and their relatives. These charts will not replace clinical judgment, but they will reduce the risk of influences from biases.

Employment outcome four years after a severe traumatic brain injury: results of the Paris severe traumatic brain injury study

OBJECTIVE

To describe employment outcome four years after a severe traumatic brain injury by the assessment of individual patients' preinjury sociodemographic data, injury-related and postinjury factors.

DESIGN

A prospective, multicenter inception cohort of 133 adult patients in the Paris area (France) who had received a severe traumatic brain injury were followed up postinjury at one and four years. Sociodemographic data, factors related to injury severity and one-year functional and cognitive outcomes were prospectively collected.

METHODS

The main outcome measure was employment status. Potential predictors of employment status were assessed by univariate and multivariate analysis.

RESULTS

At the four-year follow-up, 38% of patients were in paid employment. The following factors were independent predictors of unemployment: being unemployed or studying before traumatic brain injury, traumatic brain injury severity (i.e., a lower Glasgow Coma Scale score upon admission and a longer stay in intensive care) and a lower one-year Glasgow Outcome Scale-Extended score.

CONCLUSION

This study confirmed the low rate of long-term employment amongst patients after a severe traumatic brain injury. The results illustrated the multiple determinants of employment outcome and suggested that students who had received a traumatic brain injury were particularly likely to be unemployed, thus we propose that they may require specific support to help them find work. Implications for rehabilitation Traumatic brain injury is a leading cause of persistent disablity and can associate cognitive, emotional, physical and sensory impairments, which often result in quality-of-life reduction and job loss. Predictors of post-traumatic brain injury unemployment and job loss remains unclear in the particular population of severe traumatic brain injury patients. The present study highlights the post-traumatic brain injury student population require a close follow-up and vocational rehabilitation. The study suggests that return to work post-severe traumatic brain injury is frequently unstable and workers often experience difficulties that caregivers have to consider.

Real-time prediction of inpatient length of stay for discharge prioritization

OBJECTIVE

Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information.

MATERIALS AND METHODS

The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures.

RESULTS

The model compared to clinician predictions demonstrated significantly higher sensitivity (P < .01), lower specificity (P < .01), and a comparable Youden Index (P > .10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge.

CONCLUSIONS

There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.

Risk factors and molecular epidemiology of community-onset extended-spectrum β-lactamase-producing Escherichia coli bacteremia

PURPOSE

Inadequate empirical therapy for severe infections caused by extended-spectrum β-lactamase-producing Escherichia coli (ESBLEC) is associated with poor outcomes. This study was designed to investigate risk factors for community-onset ESBLEC bacteremia at admission to a tertiary care hospital.

MATERIALS AND METHODS

A case-control study was performed that included all episodes of ESBLEC bacteremia in the outpatient department or within 48 hours of admission from January 2005 to March 2009. Data on predisposing factors were collected. The molecular epidemiology of ESBLEC clinical isolates was also determined.

RESULTS

Among 25281 blood cultures, 60 episodes of ESBLEC bacteremia were studied, which accounted for 7% of all E. coli bacteremia at admission. Healthcare-associated infection [odds ratio (OR), 8.3; 95% confidence interval (CI), 2.4-28.7; p=0.001], malignancy (OR, 4.6; 95% CI, 1.3-16.3; p=0.018), urinary tract infection (OR, 139.1; 95% CI, 24.6-788.2; p<0.001), hepatobiliary infection (OR, 79.1; 95% CI, 13.5-463.8; p<0.001), third generation cephalosporin usage during preceding 3 months (OR, 16.4; 95% CI, 2.0-131.8; p=0.008), and severe sepsis/septic shock (OR, 73.7; 95% CI, 12.4-438.5; p<0.001) were determined as independent risk factors for community-onset ESBLEC bacteremia. The most common extended-spectrum β-lactamase (ESBL) gene identified was blaCTX-M-15 (n=31) followed by blaCTX-M-14 (n=23).

CONCLUSION

The most common types of ESBLs in E. coli causing community-onset bacteremia were CTX-M-15 and CTX-M-14 in Korea. By result of decision tree analysis, the empirical use of carbapenems is suggested only for patients with severe sepsis/septic shock, hepatobiliary infection, or healthcare-associated urinary tract infection.

A Systematic Review of Early Prognostic Factors for Return to Work After Traumatic Brain Injury

This article presents a systematic review identifying variables and their prognostic value for return to work (RTW) after traumatic brain injury (TBI). RTW has been identified as being a key goal following TBI, with estimates ranging from 10% to 70%. Prediction of postinjury employment is important for planning rehabilitation and structuring individualised vocational services. Studies examining prognostic factors were identified by searching four electronic databases, until June 2006. Searches yielded 1948 studies of which 55 met inclusion criteria and were subsequently rated for methodological quality. Mean methodological score for included studies was 3.9/6 (SD0.9, range 1–6). Analysis focused on a subset of 27 studies which provided sampling from all three domains of preinjury, injury and early postinjury variables. Few studies considered preinjury variables, apart from simple demographics. Only five studies considered preinjury employment, which was a significant predictor in each case. Severity of injury variables were invariably examined, but were significant predictors in only 8/27 studies (30%). For early postinjury variables, 14/27 studies entered cognitive variables with 12/14 (86%) identifying them as significant predictors; 3/27 studies examined neurophysical variables, with 2/3 (67%) studies finding them significant; and 12/27 studies examined multidimensional/participation variables which were statistically significant individual predictors in 8/12 (67%) cases. The results are discussed in the context of methodological issues encountered during the course of the review that require addressing in future studies.

A systematic review of factors contributing to outcomes in patients with traumatic brain injury

AIM AND OBJECTIVE

To review, systematically, factors contributing to outcomes in patients with traumatic brain injury.

BACKGROUND

Traumatic brain injury is a leading cause of death and disability. Several studies have determined the significant predictors of outcomes after traumatic brain injury. The comprehensive identification of these reliable factors for traumatic brain injury is critical to both clinical practice and research.

DESIGN

Systematic literature review.

METHODS

Eligible studies that combined at least two variables to predict outcomes in patient with traumatic brain injury were identified via electronic database searches, footnote chasing and contact with clinical experts. Quality of selected studies was assessed in terms of internal and external validity using 15 questions. Two reviewers independently examined titles, abstracts and whether each met the predefined inclusion criteria.

RESULTS

A total of 46 studies which met review criteria were finally selected. Most studies satisfied internal validity in terms of validity of research variables and multivariate analysis, but few were validated externally. The following factors were significantly associated with unfavourable outcomes: sociodemographic factors such as older age, male gender, lower level of education; clinical factors such as lower Glasgow Coma Scale score, injury caused by motor vehicle crash, hypotension, hypoxia, increased intracranial pressure, no pupil reaction, hypo- or hyperglycaemia, anaemia, coagulopathy, hypo- or hyperthermia, abnormal level of electrolytes, duration of coma; higher level of computed tomography classification by Marshall category; type of intracerebral lesions.

CONCLUSION

Further studies on integrating the sociodemographic factors, the course of the clinical condition and a unified CT scoring system, are recommended for the evaluation and improvement of the prognosis of traumatic brain injury.

RELEVANCE TO CLINICAL PRACTICE

A systematic review of factors contributing to outcome for patients with traumatic brain injury will be invaluable in triage criteria, injury prognostication, care and discharge planning, resource use and patient and family counselling.

Regional brain morphometry predicts memory rehabilitation outcome after traumatic brain injury

Cognitive deficits following traumatic brain injury (TBI) commonly include difficulties with memory, attention, and executive dysfunction. These deficits are amenable to cognitive rehabilitation, but optimally selecting rehabilitation programs for individual patients remains a challenge. Recent methods for quantifying regional brain morphometry allow for automated quantification of tissue volumes in numerous distinct brain structures. We hypothesized that such quantitative structural information could help identify individuals more or less likely to benefit from memory rehabilitation. Fifty individuals with TBI of all severities who reported having memory difficulties first underwent structural MRI scanning. They then participated in a 12 session memory rehabilitation program emphasizing internal memory strategies (I-MEMS). Primary outcome measures (HVLT, RBMT) were collected at the time of the MRI scan, immediately following therapy, and again at 1-month post-therapy. Regional brain volumes were used to predict outcome, adjusting for standard predictors (e.g., injury severity, age, education, pretest scores). We identified several brain regions that provided significant predictions of rehabilitation outcome, including the volume of the hippocampus, the lateral prefrontal cortex, the thalamus, and several subregions of the cingulate cortex. The prediction range of regional brain volumes were in some cases nearly equal in magnitude to prediction ranges provided by pretest scores on the outcome variable. We conclude that specific cerebral networks including these regions may contribute to learning during I-MEMS rehabilitation, and suggest that morphometric measures may provide substantial predictive value for rehabilitation outcome in other cognitive interventions as well.

Effects of severity of traumatic brain injury and brain reserve on cognitive-control related brain activation

Functional magnetic resonance imaging (fMRI) has revealed more extensive cognitive-control related brain activation following traumatic brain injury (TBI), but little is known about how activation varies with TBI severity. Thirty patients with moderate to severe TBI and 10 with orthopedic injury (OI) underwent fMRI at 3 months post-injury using a stimulus response compatibility task. Regression analyses indicated that lower total Glasgow Coma Scale (GCS) and GCS verbal component scores were associated with higher levels of brain activation. Brain-injured patients were also divided into three groups based upon their total GCS score (3-4, 5-8, or 9-15), and patients with a total GCS score of 8 or less produced increased, diffuse activation that included structures thought to mediate visual attention and cognitive control. The cingulate gyrus and thalamus were among the areas showing greatest increases, and this is consistent with vulnerability of these midline structures in severe, diffuse TBI. Better task performance was associated with higher activation, and there were differences in the over-activation pattern that varied with TBI severity, including greater reliance upon left-lateralized brain structures in patients with the most severe injuries. These findings suggest that over-activation is at least partially effective for improving performance and may be compensatory.

Development and validation of a risk score for post-infectious irritable bowel syndrome

OBJECTIVES

Acute gastroenteritis (GE) is an important risk factor for the development of irritable bowel syndrome (IBS). We used observational data from the Walkerton Health Study (WHS) to develop and validate a risk score for post-infectious (PI) IBS.

METHODS

Model derivation and validation were based on a split-sample method from a cohort of patients with exposure to GE (n=1,368). Study participants were randomly assigned to the derivation and validation cohorts in a 1:1 ratio. Within the derivation cohort, univariate and multivariable logistic regression were used to identify risk factors associated with IBS. The risk model was then applied to the validation cohort. Overall model performance was assessed using the area under the receiver operating curve (ROC). The risk score was developed using multivariable regression coefficients obtained from the derivation set and validated in the validation set. Classification and regression tree (CART) modeling was used to determine cutoff values for high, intermediate, and low risk based on the total score.

RESULTS

Nine variables were identified as important predictors of IBS (gender, age<60, longer duration of diarrhea, increased stool frequency, abdominal cramping, bloody stools, weight loss, fever, and psychological disorders (anxiety and depression)). The discriminatory power of the risk model based on the area under ROC was 0.70 and was similar in the validation set. The risk score model showed good accuracy in both the derivation and validation sets and was able to distinguish among cohorts at low, intermediate, and high risk for developing PI-IBS. Percentages of patients with PI-IBS in the low, intermediate and high risk were 10, 35, and 60% in the derivation cohort and 17, 36, and 62% in the validation cohort.

CONCLUSIONS

A simple risk tool that uses demographics and symptoms of acute GE can predict which patients with acute GE are at risk of developing PI-IBS. This tool may be used clinically to assess risk and to guide treatment.

Return to productivity following traumatic brain injury: Cognitive, psychological, physical, spiritual, and environmental correlates

PURPOSE

The purpose of this study was to investigate the determinants and correlates of return to productivity (RTP) defined here as return to paid employment and/or school four years following traumatic brain injury (TBI).

METHOD

Participants included 46 people with TBI, part of a prospective, cohort study, and 14 friend/family member controls all employed and/or in school at time of injury or inception into the study. Variables were selected for investigation based on two models of recovery. Demographic and injury severity data including time to recover free recall were collected at time of injury, on admission to a trauma unit. Data on other variables (neuropsychological, psychological, physical, spiritual, environmental) were collected concurrent with productivity status at a mean of 4.3 years post-TBI.

RESULTS

Time to recover free recall (measured acutely), neuropsychological status, pain severity, depression, and the use of maladaptive coping behaviours were all related to productivity status (p < 0.05). When these variables were entered into exploratory, planned hierarchical logistic regression models time to free recall, pain, and maladaptive coping remained in the models with depression only dropping out because of the high correlation with pain (r > 0.80).

CONCLUSIONS

Injury severity (time to free recall), physical status (pain), and psychological status (depression, coping) are important to understanding differences in productivity outcomes. Addressing pain, depression and coping in rehabilitation programs may have a positive impact on outcomes.

Development and validation of a colon cancer risk assessment tool for patients undergoing colonoscopy

OBJECTIVES

Diagnostic criteria for hereditary colorectal cancer (CRC) are complex. "Open-access" colonoscopy makes it challenging to identify who needs genetic evaluation, intensive surveillance, and screening for extracolonic tumors. Our aim was to develop a simple, preprocedural risk assessment tool to identify who may be at highest risk for CRC.

METHODS

A total of 631 outpatients undergoing colonoscopy at two academic practices completed a questionnaire assessing personal and family histories of CRC, polyps, and Lynch syndrome (LS)-associated malignancies. Subjects were considered to be high-risk if one of the nine prespecified characteristics of hereditary CRC syndromes was met. Through recursive partitioning analysis, an algorithm of fewest questions needed to capture the most high-risk individuals was developed. The results were validated in 5,335 individuals undergoing colonoscopy at five private endoscopy centers and tested in 285 carriers of mismatch repair mutations associated with LS.

RESULTS

About 17.7% and 20.0% of individuals were classified as high-risk in the development and validation cohorts, respectively. Recursive partitioning revealed three questions that were most informative for identifying high-risk patients: (i) "Do you have a first-degree relative with CRC or LS-related cancer diagnosed before age 50?" (ii) "Have you had CRC or polyps diagnosed before age 50?" (iii) "Do you have > or =3 relatives with CRC?" When asked successively, these questions identified 77% of high-risk individuals in both cohorts and 271 of 285 (95%) of mutation carriers.

CONCLUSIONS

Approximately one in five individuals undergoing colonoscopy would benefit from further risk assessment. We developed a simple, three-question CRC Risk Assessment Tool to identify the majority of patients who require additional assessment and possible genetic evaluation.

Prediction of memory rehabilitation outcomes in traumatic brain injury by using functional magnetic resonance imaging

OBJECTIVE

To evaluate the ability of functional magnetic resonance imaging (fMRI) measures collected from people with traumatic brain injury (TBI) to provide predictive value for rehabilitation outcomes over and above standard predictors.

DESIGN

Prospective study.

SETTING

Academic medical center.

PARTICIPANTS

Persons (N=54) with TBI greater than 1 year postinjury.

INTERVENTION

A novel 12-session group rehabilitation program focusing on internal strategies to improve memory.

MAIN OUTCOME MEASURE

The Hopkins Verbal Learning Test-Revised (HVLT-R) delayed recall score.

RESULTS

fMRI measures were collected while participants performed a strategically directed word memorization task. Prediction models were multiple linear regressions with the following primary predictors of outcome: age, education, injury severity, preintervention HVLT-R, and task-related fMRI activation of the left dorsolateral and left ventrolateral prefrontal cortex (VLPFC). Baseline HVLT-R was a significant predictor of outcome (P=.007), as was injury severity (for severe vs mild, P=.049). We also found a significant quadratic (inverted-U) effect of fMRI in the VLPFC (P=.007).

CONCLUSIONS

This study supports previous evidence that left prefrontal activity is related to strategic verbal learning, and the magnitude of this activation predicted success in response to cognitive memory rehabilitation strategies. Extreme under- or overactivation of VLPFC was associated with less successful learning after rehabilitation. Further study is necessary to clarify this relationship and to expand and optimize the possible uses of functional imaging to guide rehabilitation therapies.

Prognostic factors of long-term functioning and productivity after traumatic brain injury: a systematic review of prospective cohort studies

OBJECTIVE

To systematically review prospective cohort studies that investigated prognostic factors associated with long-term activity limitations or participation restrictions and productivity after a traumatic brain injury.

DATA SOURCES

PubMed and Psychinfo were searched from 1995 to April 2005, and references were checked.

REVIEW METHODS

Publications were selected if the study assessed prognostic factors for activity limitations or participation restrictions at least one year post injury; outcome was measured with another or additional measure besides the Glasgow Outcome Scale; the design was a prospective cohort study of adult traumatic brain injury patients; the article was a full-text article written in English, French, German or Dutch. Two reviewers independently assessed methodological quality. A study was considered as 'high quality' if it satisfied at least half of the maximum available quality score.

RESULTS

Thirty-five articles reporting on 14 cohorts were included. Due to heterogeneity in prognostic factors and outcome measures, a best-evidence synthesis was performed. All cohorts were of high quality. Strong evidence for predicting disability was found for older age, pre-injury unemployment, pre-injury substance abuse, and more disability at rehabilitation discharge. Strong prognostic factors for being non-productive were pre-injury unemployment, longer post-traumatic amnesia, more disability at rehabilitation admission, and pre-injury substance abuse.

CONCLUSION

Older age, pre-injury unemployment, pre-injury substance abuse and more disability at rehabilitation discharge are important predictors of long-term disability. Pre-injury unemployment, longer post-traumatic amnesia, more disability at rehabilitation admission and pre-injury substance abuse are important predictors of being non-productive.

New approaches to physiological informatics in neurocritical care

INTRODUCTION

A fundamental purpose of neurocritical care is the management of secondary brain injury. This is often accomplished by monitoring and managing individual patient parameters including physiological vital signs. Yet, the ability to record physiological data exceeds our ability to fully integrate it into patient care. We propose that advances in monitoring must be accompanied by advances in methods of high-frequency, multivariate data analysis that integrate the multiple processes occurring in critically ill patients.

METHODS

We describe initial work in the emerging field of physiological informatics in critical care medicine. We analyzed data on 23 patients with brain injury from our Neurotrauma and Critical Care Database, which contains more than 20 physiological parameters recorded automatically at one-minute intervals via bedside monitors connected to standard personal computers. We performed exploratory data analysis, studied two patient cases in detail, and implemented a data-driven classification approach using hierarchical clustering.

RESULTS

In this study, we present challenges and opportunities for high-frequency multimodal monitoring to quantitatively detect secondary brain insults, and develop clustering methodology to construct multivariate physiological data "profiles" to classify patients for diagnosis and treatment.

CONCLUSIONS

Recording of many physiological variables across multiple patients is feasible and can lead to new clinical insights. Computational and analytical methods previously used primarily for basic science may have clinical relevance and can potentially be adapted to provide physicians with improved ability to integrate complex information for decision making in neurocritical care.

Executive function and coping at one-year post traumatic brain injury

The purpose of this study was to examine the relationship between executive function and coping at one-year-post traumatic brain injury (TBI). TBI and matched control groups completed a coping questionnaire and a neuropsychological test series. In the TBI group, better executive performance was related to the use of problem focused coping (considered more adaptive). Conversely, lower executive performance was related to the use of emotion focused coping (considered more maladaptive). Planned hierarchical regression showed that executive function contributed significantly to the use of problem focused coping above and beyond pre-morbid intelligence and injury severity. Implications for cognitive rehabilitation are discussed.

The Accuracy of Artificial Neural Networks in Predicting Long-term Outcome After Traumatic Brain Injury

OBJECTIVE

This study compared the accuracy of artificial neural networks to multiple regression and classification and regression trees in predicting outcomes of 1,644 patients in the Traumatic Brain Injury Model Systems database 1 year after injury.

METHODS

Data from rehabilitation admission were used to predict discharge scores on the Functional Independence Measure, the Disability Rating Scale, and the Community Integration Questionnaire.

RESULTS

Artificial neural networks did not demonstrate greater accuracy in predicting outcomes than did the more widely used method of multiple regression. Both of these methods outperformed classification and regression trees.

CONCLUSION

Because of the sophisticated form of multiple regression with splines that was used, firm conclusions are limited about the relative accuracy of artificial neural networks compared to more widely used forms of multiple regression.

A quick and reliable screening measure for OCD in youth: reliability and validity of the obsessive compulsive scale of the Child Behavior Checklist

BACKGROUND

The high prevalence and morbidity of obsessive compulsive disorder (OCD) in youth, the secretive nature of the disorder leading to under-recognition, and the lack of specialized child psychiatry services in many areas suggest that a simple, quick, and reliable screening tool to identify cases could be very useful to clinicians who work with children.

METHOD

We used 8 items from the Child Behavior Checklist (CBCL), an empirically derived instrument free of clinician bias, to investigate the usefulness of a previously reported CBCL-based obsessive compulsive scale (OCS) by Nelson et al [Nelson EC, Hanna GL, Hudziak JJ, Botteron KN, Heath AC, Todd RD. Obsessive-compulsive scale of the Child Behavior Checklist: Specificity, sensitivity, and predictive power. Pediatrics 2001;108(1):E14] in a separate cohort of youth with OCD. We computed the psychometric properties of the OCS in our sample of youth with OCD and in psychiatric and normal controls, and compared these to the published values.

RESULTS

Using the recommended cutoff between the 60th and 70th percentiles of the OCS to best predict the presence of OCD, we found very high sensitivity (92%-78%), specificity (86%-94%), negative predictive value (96%-90%), and positive predictive value (77%-86%).

CONCLUSIONS

The OC scale of the CBCL shows good reliability and validity and acceptable psychometric properties to help discriminate youth with OCD.

A description of morbidly obese state employees requesting a bariatric operation

BACKGROUND

The federal government, the medical insurance industry, and the academic medical community have disagreed over what treatments are appropriate and cost effective for morbid obesity. This debate is hindered by inadequate data regarding the true costs of diseases and who chooses an operation as a treatment option. The purpose of this study was to obtain these costs and to describe this population.

METHODS

Louisiana's managed medical insurance program created primarily for its civil service employees contracted to offer a small random group of morbidly obese employees the option of a bariatric operation. This observational study examined the subpopulation who requested consideration for the operation. We present historic cost data from all medical expenses paid by the insurance company, a telephone survey of the volunteers in the study to determine their medical problems, and diagnostic evaluation data on those employees randomized to proceed for possible bariatric operation.

RESULTS

A total of 911 of 189,398 adult members of the insurance plan wanted to be considered for this study. Only 397, however, completed the informed-consent process. Of the 248 employees who met the age requirement, body mass index criteria, and health criteria to be considered for a bariatric operation and were randomized, 20 withdrew before obtaining 40 committed operative candidates. The 773 morbidly obese female members had used a mean of dollar 11,145 in medical insurance expenses in the year 2003 versus a mean of dollar 8,096 for the other 106,908 adult women. Similar values for the men were dollar 16,720 for the 138 morbidly obese men versus dollar 5,943 for the other 82,490 men.

CONCLUSIONS

The morbidly obese members of this medical insurance plan who requested a bariatric operation are costing their plan 1.4 to 2.8 times the yearly amount of the other adult members in medical expenses. The yearly mean amount the insurance plan spends on these members suggests that operative treatment would pay for itself in a relatively few number of years if it could significantly reduce these costs. Even in those who consider bariatric operation, many withdraw, further limiting the costs of operative therapy.

Workers' risk of unemployment after traumatic brain injury: a normed comparison

We examined, among those persons working preinjury, the risk of unemployment 1 year after traumatic brain injury (TBI) relative to expected risk of unemployment for the sample under a validated risk-adjusted econometric model of employment in the U.S. population. Results indicate that 42% of TBI cases were unemployed versus 9% expected, relative risk (RR) = 4.5, 95% confidence interval (CI) (4.12, 4.95). The relative risk for unemployment was higher among males, those with higher education, persons with more severe injuries, and more impaired early neuropsychological or functional status. Difference in unemployment rates gave similar results for gender, severity of injury, and early neuropsychological and functional status. However, for education, the excess was smaller among those more highly educated, but the unemployment rate in the more highly educated in the general population was sufficiently small to yield a larger relative risk. In conclusion, after accounting for underlying risk of unemployment in the general population, unemployment is substantially higher after TBI for people who were employed when they were injured. The differential employment status varies depending on demographics, severity of brain injury, early functional outcome, and neurobehavioral indicators. For characteristics such as education, associated with rates of unemployment in the general population, different methods used to compare the rates may yield different results.

Childhood antecedents to panic disorder in referred and nonreferred adults

OBJECTIVE

We used a recursive partitioning method to examine antecedent childhood anxiety disorders in large samples of referred and nonreferred subjects with and without panic disorder.

METHODS

Referred subjects included adults treated for panic disorder (n = 131) and comparison adults with neither major anxiety nor mood disorders (n = 61). The nonreferred adult group derived from an opportunistic sample originally ascertained through family studies of probands with and without attention-deficit/hyperactivity disorder (ADHD), yielding 58 adults with panic disorder and 587 who were free of major anxiety and mood disorders.

RESULTS

The majority of referred (65%) and nonreferred (52%) adults with panic disorder had antecedent childhood anxiety or disruptive behavior disorders. Classification and Regression Trees (CART) analysis showed that both separation anxiety disorder and overanxious disorder were independent predictors of subsequent panic disorder in both referred and nonreferred samples.

CONCLUSIONS

These results confirm and extend previously reported findings by documenting that childhood anxiety disorders are important antecedent risk factors for panic disorder, independently of referral bias.

A case study found that a regression tree outperformed multiple linear regression in predicting the relationship between impairments and Social and Productive Activities scores

OBJECTIVE

Many important physiologic and clinical predictors are continuous. Clinical investigators and epidemiologists' interest in these predictors lies, in part, in the risk they pose for adverse outcomes, which may be continuous as well. The relationship between continuous predictors and a continuous outcome may be complex and difficult to interpret. Therefore, methods to detect levels of a predictor variable that predict the outcome and determine the threshold for clinical intervention would provide a beneficial tool for clinical investigators and epidemiologists.

STUDY DESIGN AND SETTING

We present a case study using regression tree methodology to predict Social and Productive Activities score at 3 years using five modifiable impairments. The predictive ability of regression tree methodology was compared with multiple linear regression using two independent data sets, one for development and one for validation.

RESULTS

The regression tree approach and the multiple linear regression model provided similar fit (model deviances) on the development cohort. In the validation cohort, the deviance of the multiple linear regression model was 31% greater than the regression tree approach.

CONCLUSION

Regression tree analysis developed a better model of impairments predicting Social and Productive Activities score that may be more easily applied in research settings than multiple linear regression alone.

Repeated split sample validation to assess logistic regression and recursive partitioning: an application to the prediction of cognitive impairment

Screening strategies play an important part in the identification and diagnosis of illness. Testing of such strategies in a clinical trial can have important implications for the treatment of such illnesses. Before the clinical trial, however, it is important to develop a practical screening/classification procedure that accurately predicts the presence of the illness in question. Recent published studies have shown a growing preference for classification tree/recursive partitioning procedures.This paper compares the application of logistic regression and recursive partitioning to a neuropsychological data set of 252 patients recruited from four Veterans Affairs Medical Centers. Logistic regression and recursive partitioning was used to predict cognitive impairment in 12 randomly selected exploratory/validation samples. We assessed the effect of sampling on variable selection and predictive accuracy.Predictive accuracy of the logistic regression and recursive partitioning procedures was comparable across the exploratory data samples but varied across the validation samples. Based on shrinkage, both classification procedures performed equally well for the prediction of cognitive impairment across the twelve samples. While logistic regression provided an estimated probability of outcome for each patient, it required several mathematical calculations to do so. However, logistic regression selected one or two less predictors than recursive partitioning with comparable predictive accuracy. Recursive partitioning, on the other hand, readily identified patient characteristics and variable interactions, was easy to interpret clinically and required no mathematical calculations. There was a high degree of overlap of the predictor variables between the two procedures.In the context of neuropsychological screening, logistic regression and recursive partitioning performed equally well and were quite stable in the selection of predictors for the identification of patients with cognitive impairment, although recursive partitioning may be easier to use in a clinical setting because it is based on a simple decision tree.

Data mining tools for biological sequences

We describe a methodology, as well as some related data mining tools, for analyzing sequence data. The methodology comprises three steps: (a) generating candidate features from the sequences, (b) selecting relevant features from the candidates, and (c) integrating the selected features to build a system to recognize specific properties in sequence data. We also give relevant techniques for each of these three steps. For generating candidate features, we present various types of features based on the idea of k-grams. For selecting relevant features, we discuss signal-to-noise, t-statistics, and entropy measures, as well as a correlation-based feature selection method. For integrating selected features, we use machine learning methods, including C4.5, SVM, and Naive Bayes. We illustrate this methodology on the problem of recognizing translation initiation sites. We discuss how to generate and select features that are useful for understanding the distinction between ATG sites that are translation initiation sites and those that are not. We also discuss how to use such features to build reliable systems for recognizing translation initiation sites in DNA sequences.

Cognitive Neuroactivation Using SPECT and the Stroop Colored Word Test in Patients with Diffuse Brain Injury

Psychomotor slowing in patients with diffuse brain injury frequently underlies impaired cognitive performance on neuropsychological tests, for example, the Stroop Colored Word test. The aim of the present study was to determine the neural basis associated with performance on the Stroop interference subtask in patients with diffuse brain injury. We hypothesized that patients would be slower than healthy controls, and that this would be associated with brain activations other than those seen in healthy subjects. Brain perfusion, using a split-dose activation paradigm with single photon emission tomography (SPECT) and the Stroop test, was assessed in 9 patients with diffuse brain injury. The Stroop interference score was calculated as a behavioral parameter, and functional imaging data were analyzed with statistical parametrical mapping (SPM99) to determine significant voxel-wise differences of activation between the control and the activation condition. Patients were impaired on the interference subtask of the Stroop test. Comparison of the SPECT data obtained during the activation condition with those obtained during the control condition by means of SPM showed significant activations in the left inferior parietal lobe, the right anterior cingulate extending into the right middle frontal gyrus and the right caudate, and the left posterior cingulate cortex. Patients with diffuse brain injury were slower than healthy controls on the interference subtask of the Stroop test, suggesting difficulty with resistance to distractions. This finding was associated with activation effects in posterior (mainly parietal) brain areas in addition with activation of previously observed anterior (mainly anterior cingulate) brain regions.

Classification and Regression Tree for Prediction of Outcome after Severe Head Injury Using Simple Clinical and Laboratory Variables

Many previous studies have constructed several predictive models for outcome after severe head injury, but these have often used expensive, time consuming, or highly specialized measurements. The goal of this study was to develop a simple, easy to use a model involving only variables that are rapidly and easily achievable in daily routine practice. To this end, a classification and regression tree (CART) technique was employed in the analysis of data from 345 patients with isolated severe brain injury who were admitted to Asclepeion General Hospital of Athens from January, 1993, to December, 2000. A total of 16 prognostic indicators were examined to predict neurological outcome at 6 months after head injury. Our results indicated that Glasgow Coma Scale was the best predictor of outcome. With regard to the other data, not only the most widely examined variables such as age, pupillary reactivity, or computed tomographic findings proved again to be strong predictors, but less commonly applied parameters, indirectly associated with brain damage, such as hyperglycemia and leukocytosis, were found to correlate significantly with prognosis too. The overall cross-validated predictive accuracy of CART model for these data was 86.84%, with a cross-validated relative error of 0.308. All variables included in this tree have been shown previously to be related to outcome. Methodologically, however, CART is quite different from the more commonly used statistical methods, with the primary benefit of illustrating the important prognostic variables as related to outcome. This technique may prove useful in developing new therapeutic strategies and approaches for patients with severe brain injury.

Classification and regression tree analysis in public health: methodological review and comparison with logistic regression

BACKGROUND

Audience segmentation strategies are of increasing interest to public health professionals who wish to identify easily defined, mutually exclusive population subgroups whose members share similar characteristics that help determine participation in a health-related behavior as a basis for targeted interventions. Classification and regression tree (C&RT) analysis is a nonparametric decision tree methodology that has the ability to efficiently segment populations into meaningful subgroups. However, it is not commonly used in public health.

PURPOSE

This study provides a methodological overview of C&RT analysis for persons unfamiliar with the procedure.

METHODS AND RESULTS

An example of a C&RT analysis is provided and interpretation of results is discussed. Results are validated with those obtained from a logistic regression model that was created to replicate the C&RT findings. Results obtained from the example C&RT analysis are also compared to those obtained from a common approach to logistic regression, the stepwise selection procedure. Issues to consider when deciding whether to use C&RT are discussed, and situations in which C&RT may and may not be beneficial are described.

CONCLUSIONS

C&RT is a promising research tool for the identification of at-risk populations in public health research and outreach.

Nonrandomized studies of rehabilitation for traumatic brain injury: can they determine effectiveness?

OBJECTIVE

To examine the feasibility of investigating rehabilitation effectiveness for traumatic brain injury (TBI) with a nonrandomized design.

DESIGN

Observational cohort with confounder control by regression methodology.

SETTING

Level I trauma center.

PARTICIPANTS

Consecutive series of 365 individuals with TBI discharged to inpatient rehabilitation or home (78% follow-up).

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

The Glasgow Outcome Scale (GOS), Sickness Impact Profile (SIP), Burden Inventory, and Perceived Quality of Life (PQOL). The predictors of interest: discharge to comprehensive inpatient rehabilitation or home and inpatient rehabilitation length of stay (LOS).

RESULTS

Discharge to rehabilitation was associated with poorer functioning on the GOS (P=.03) and SIP (P=.57), an increase on the Burden Inventory (P=.14), and improved PQOL (P=.20). Similar results were found for longer lengths of inpatient rehabilitation.

CONCLUSIONS

The results appear to be because of a confounding effect rather than rehabilitation. The study design could not control for confounding that resulted from unmeasured or difficult to measure aspects of the clinical decisions for discharge placement and rehabilitation LOS. Furthermore, typical severity indices were inadequate to control for injury severity and recovery. Matching designs that investigate TBI rehabilitation are also at risk for inadequate confounder control.

Prediction of fracture in nursing home residents

OBJECTIVES

To investigate cross-validated methods of identifying patients at increased risk of fracture in nursing homes using readily available data.

DESIGN

Prospective cohort study with 18 months of follow-up.

SETTING

Forty-seven randomly selected nursing homes in Maryland.

PARTICIPANTS

One thousand four hundred twenty-seven white female nursing home residents aged 65 and older were followed for fracture for 18 months after baseline assessment.

MEASUREMENTS

Fracture ascertained by physician note or x-ray from chart abstraction; demographic and baseline data extracted from the Minimum Data Set (MDS).

RESULTS

Exploratory analyses on a random subset (67%) of the data (development sample) identified variables that might be important in predicting subsequent fracture and included variables for how the resident moved between locations in her room or adjacent corridor (mobility), age, weight, height, independence in eating and dressing, urinary incontinence, resistance to care, falls in the previous 6 months, a dementia score, and other activities of daily living. A simple scoring algorithm derived from a subset of these MDS variables showed good sensitivity (.70) but low specificity (.39) in the random validation sample.

CONCLUSION

A scoring algorithm developed in more than 1,400 white females from 47 nursing homes in the state of Maryland shows high sensitivity for identifying women at increased risk for fracture and may be useful in targeting fracture prevention programs.

Prediction of functional outcomes after traumatic brain injury: a comparison of 2 measures of duration of unconsciousness

OBJECTIVE

To compare the usefulness of time until motor localization occurs versus time until commands are followed in predicting outcome after traumatic brain injury (TBI).

DESIGN

A retrospective analysis of data from a prospective cohort study of subjects with severe TBI.

SETTING

Seventeen Traumatic Brain Injury Model System programs.

PARTICIPANTS

A total of 496 subjects, recruited through the TBI Model System programs, with loss of consciousness greater than 1 day, with no late neurosurgical complications, and complete data for all measures.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Time until Glasgow Coma Scale (GCS) motor score of 5 (time to motor localization) and time until GCS motor score of 6 (time until commands were followed) were abstracted from medical records. Functional outcomes were assessed at inpatient rehabilitation admission and discharge, along with acute and rehabilitation lengths of stay and charges.

RESULTS

Time until commands were followed was a better predictor of all of the outcomes assessed than time until motor localization occurred. In multiple regression models, time until motor localization did not add significantly to the prediction provided by time until commands were followed. The predictive power of time to command following was superior even in the subgroup with poor language comprehension as measured by the Token Test.

CONCLUSION

Despite the theoretical appeal of time to motor localization (eg, in persons with language comprehension problems), time to command following appears to be a more powerful predictor of outcome after severe brain injury.

Early predictors of mortality and morbidity after severe closed head injury

Mortality and morbidity of 158 patients with severe head injury were studied in relation to age, and early (24-h) clinical and computed tomography data. For comparison of outcome data in survivors, a group of 32 patients with traumatic injuries to parts of the body other than the head was used as controls. Within the head-injured group, the mortality rate was 51%. Logistic regression analyses combined 13 out of 16 predictors into a model with an accuracy of 93%, a sensitivity of 90%, and a specificity of 95%. These include age, Glasgow Coma Scale (GCS) score, pupillary reactivity, blood pressure, intracranial pressure, blood glucose, platelet count, body temperature, cerebral lactate, and subdural, intracranial, subarachnoid, and ventricular hemorrhage. At 6 months postinjury, head-injury survivors and trauma controls were evaluated with the Glasgow Outcome Scale (GOS), a neuropsychological test battery and the Sickness Impact Profile (SIP). Head-injury survivors had a higher proportion of disabilities and neuropsychological dysfunctions than trauma controls. They also report more quality of life-related functional limitations on the SIP scales for mobility, intellectual behavior, communication, home management, eating, and work. Linear regression analysis resulted in age being the only important predictor of outcome on the GOS, the GCS score being the best predictor of neuropsychological functioning, and pupillary reactivity being the most predictive for self-reported quality of life as measured by SIP. Those factors important for predicting mortality (clinical variables such as ICP or blood glucose level, and CT observations) failed to show any significant relationship with morbidity.

Predicting elementary school participation in children with disabilities

OBJECTIVE

To identify predictors of participation in school activities from two sets of functional variables using classification and regression tree analysis.

DESIGN

Relational study.

PARTICIPANTS

A nationwide sample of 341 children with various disabling conditions, including physical and cognitive/behavioral types of impairment and various severity levels. Children attended public elementary school in 40 states in the United States.

MAIN OUTCOME MEASURE

Overall participation in elementary school, combining children's participation in six different environments (transportation, transitions, classroom, cafeteria, bathroom, and playground), as measured by the newly developed School Function Assessment. The children were dichotomized into full (n = 117) and limited (n = 224) participation categories.

RESULTS

Two classification trees were developed identifying a small set of predictors from variables measuring performance of functional tasks and discrete activities. Final predictive models included physical and cognitive-behavioral variables, suggested important interactions among predictors, and identified meaningful cut-off points that classified the sample into the outcome categories with about 85% accuracy.

CONCLUSIONS

Limited participation was predicted by information about children's physical capabilities. Full participation was predicted by a combination of physical and cognitive-behavioral variables. Findings underscore the relative utility of functional performance compared with impairment information to predict the outcome, and suggest pathways of influence to consider in future research and intervention efforts.

Should the ventilator be withdrawn? Attitudes of the general public, nurses and physicians

In a Swedish nation-wide study, 1196 members of the general public, and 121 physicians and 339 nurses drawn from 29 intensive care units were questioned about the use of ventilator treatment for severely ill patients. Response rates were 64%, 88% and 86%, respectively. Two typical case history scenarios were presented: one describing a conscious and competent patient with pneumonia and severe cancer, and the other describing a patient who had been comatose for 1 month following head trauma. In the case of the cancer patient, 49% of the general public, 63% of the physicians and 59% of the nurses answered that they would wish that the ventilator treatment be discontinued, if they were the patient. In the case of the comatose patient, 48% of the general public, 82% of the physicians and 70% of the nurses answered that they would wish that the ventilator treatment to be discontinued, if they were a relative of the patient. Respondents own preferences, in the three groups, for life support favoured withdrawal of ventilator treatment.

Predicting survival using simple clinical variables: a case study in traumatic brain injury

OBJECTIVES

Prediction of patient outcome can be useful as an aid to clinical decision making, to explore possible biological mechanisms, and as part of the clinical audit process. Many studies have constructed predictive models for survival after traumatic brain injury, but these have often used expensive, time consuming, or highly specialised measurements. The aim of this study was to develop a simple easy to use model involving only variables which are rapidly and easily clinically achievable in routine practice.

METHODS

All consecutive patients admitted to a regional trauma centre with moderate or severe head injury were enrolled in the study. Basic demographic, injury, and CT characteristics were recorded. Patient survival at 1 year was used to construct a simple predictive model which was then validated on a very similar patient group.

RESULTS

372 patients were included in the study, of whom 365 (98%) were followed up for survival at 1 year. Multiple logistic regression resulted in a model containing age (p<0.001), Glasgow coma scale score (p<0.001), injury severity score (p<0.001), pupil reactivity (p=0.004), and presence of haematoma on CT (p=0.004) as independently significant predictors of survival. The model was validated on an independent set of 520 patients, showing good discrimination and adequate calibration, but with a tendency to be pessimistic about very severely injured patients. It is presented as an easy to use nomogram.

CONCLUSIONS

All five variables have previously been shown to be related to survival. All variables in the model are clinically simple and easy to measure rapidly in a centre with access to 24 hour CT, resulting in a model that is both well validated and clinically useful.

Mild traumatic lesion of the right parietal cortex of the rat: selective behavioural deficits in the absence of neurological impairment

Fluid impact models are widely used to study the histological and neurochemical consequences of traumatic brain injury and although behavioural consequences have also been studied, behavioural changes are often confounded by non-specific neurological deficits. In the present study we investigated behavioural effects of a unilateral mild traumatic lesion of the right lateral parietal cortex. This region is implicated in a number of basic and complex behaviors, and we therefore analyzed the performance of rats in a diverse range of behavioural procedures. The lesion had no effects on general neurological function, motor activity (activity boxes, rota-rod and paw reaching tests), habituation to a novel environment (holeboard), spatial learning ability (Morris water maze) or anxiety (elevated plus-maze). However, the lesioned animals demonstrated lower levels of exploration than the control group when novel objects were placed beneath some of the holes in the holeboard. Lesioned animals also differed from controls in their performance in passive and active avoidance procedures. In a step-through passive avoidance test the lesioned rats performed worse than the sham-operated controls, i.e. they had significantly lower entry latencies on the 2nd day. In contrast, in the active avoidance task the lesioned animals performed better than sham-operated rats, demonstrating a better ability to learn to avoid and escape from the shock. These diverse results in different tests of learning and memory, in particular the impairment in passive avoidance and the improvement in active avoidance behavior, are difficult to reconcile with a simple effect of the lesion on cognitive performance per se. The complete absence of general neurological deficits following the mild traumatic injury rules out the possibility that the observed behavioural changes reflect a non-specific impairment. These results demonstrate that mild traumatic lesion of the right parietal cortex can induce relatively selective behavioural changes that may serve to study functional recovery after trauma. However further work is required to establish the underlying deficit(s) that has led to the behavioural effects described here.