Effects of service delivery versus changes in incidence on trends in injury: a demonstration using hospitalised traumatic brain injury

Though injury incidence and hospitalisations are likely to be correlated, a range of factors other than incidence of injury in a population may influence trends in hospitalised injuries. These include technical changes in the hospital data's coverage, and real changes in the incidence of hospitalisations independent of the population incidence. This paper addresses the latter using the example of traumatic brain injury (TBI) hospitalisations in New Zealand.Data were New Zealand public hospitals inpatient discharges. Five measures of TBI severity were used.The relative rate of minor to serious TBI hospitalisations declined by approximately 2-6% per year from 1988 to 1998. This decline is observed across different mechanisms and the two severity measures used in the detailed analysis.The relative decline in minor to serious TBI is likely to be related to a change in the probability of admission rather than a change in the population TBI incidence. As most TBI hospitalisations are minor this suggests the trend in TBI hospitalisations was significantly influenced by factors other than changes in population incidence. Any analysis of routinely collected secondary injury data needs to consider case selection carefully, especially if trends are being examined. Applying a severity threshold should give more reliable trends.     

Re‐evaluation of re‐hospitalization and rehabilitation in renal research

Introduction: The use of administrative data to capture 30‐day readmission rates in end‐stage renal disease is challenging since Medicare combines claims from acute care, inpatient rehabilitation (IRF), and long‐term care hospital stays into a single “Inpatient” file. For data prior to 2012, the United States Renal Data System does not contain the variables necessary to easily identify different facility types, making it likely that prior studies have inaccurately estimated 30‐day readmission rates. Methods: For this report, we developed two methods (a “simple method” and a “rehabilitation‐adjusted method”) to identify acute care, IRF, and long‐term care hospital stays from United States Renal Data System claims data, and compared them to methods used in previously published reports. Findings: We found that prior methods overestimated 30‐day readmission rates by up to 12.3% and overestimated average 30‐day readmission costs by up to 11%. In contrast, the simple and rehabilitation‐adjusted methods overestimated 30‐day readmission rates by 0.1% and average 30‐day readmission costs by 1.8%. The rehabilitation‐adjusted method also accurately identified 96.8% of IRF stays. Discussion: Prior research has likely provided inaccurate estimates of 30‐day readmissions in patients undergoing dialysis. In the absence of data on specific facility types particularly when using data prior to 2012, future researchers could employ our method to more accurately characterize 30‐day readmission rates and associated outcomes in patients with end‐stage renal disease.     

Crash risk factors related to individuals sustaining and drivers following traumatic brain injuries

Objective

Some crashes result in drivers experiencing (or sustaining) a traumatic brain injury (TBI) while other crashes involve drivers that have already experienced a TBI. The objective of this study is to examine the factors that influence these two TBI crash groups.

Methods

Data from the Iowa Department of Public Health's Brain Injury Registry and Department of Transportation's crash records were linked together and used in logistic regression models to predict the likelihood of a driver sustaining a TBI in a crash and those who drive after a TBI.

Results

Between 2001 and 2006, there were 2382 crashes in which an individual sustained a TBI. As expected, a higher likelihood of sustaining a TBI was observed for motorcycle drivers who did not wear a helmet and in crashes that resulted in total or disabling vehicle damage. Focusing specifically on the post-TBI drivers (and not occupants), 1583 were involved in crashes. These post-TBI drivers were less likely to wear seatbelts or have passengers in the vehicle at the time of the crash, and were more likely to crash at night. Post-TBI drivers were also involved in significantly more multiple crashes (about 14%) when compared to drivers who have not experienced a TBI (about 10%) during the study period. When controlling for gender, date of injury, and severity of TBI (using Glasgow Coma Scale), individuals that sustained a TBI when they were younger were more likely to be involved in multiple crashes.

Conclusions

Different factors influence the crash likelihood for those that sustain a TBI in a crash and those that crash following a TBI. In general, post-TBI drivers have a higher occurrence of multiple crashes and this should be further explored to guide driver rehabilitation, evaluation, and training.     

Internet-based monitoring and benchmarking in ambulatory surgery centers

BACKGROUND: Each year the number of surgical procedures performed on an outpatient basis increases, yet relatively little is known about assessing and improving quality of care in ambulatory surgery. Conventional methods for evaluating outcomes, which are based on assessment of inpatient services, are inadequate in the rapidly changing, geographically dispersed field of ambulatory surgery. Internet-based systems for improving outcomes and establishing benchmarks may be feasible and timely. METHODS: Eleven freestanding ambulatory surgery centers (ASCs) reported process and outcome data for 3,966 outpatient surgical procedures to an outcomes monitoring system (OMS), during a demonstration period from April 1997 to April 1999. ASCs downloaded software and protocol manuals from the OMS Web site. Centers securely submitted clinical information on perioperative process and outcome measures and postoperative patient telephone interviews. Feedback to centers ranged from current and historical rates of surgical and postsurgical complications to patient satisfaction and the adequacy of postsurgical pain relief. RESULTS: ASCs were able to successfully implement the data collection protocols and transmit data to the OMS. Data security efforts were successful in preventing the transmission of patient identifiers. Feedback reports to ASCs were used to institute changes in ASC staffing, patient care, and patient education, as well as for accreditation and marketing. The demonstration also pointed out shortcomings in the OMS, such as the need to simplify hardware and software installation as well as data collection and transfer methods, which have been addressed in subsequent OMS versions. DISCUSSION: Internet-based benchmarking for geographically dispersed outpatient health care facilities, such as ASCs, is feasible and likely to play a major role in this effort.     

Outcomes following physical restraint reduction programs in two acute care hospitals.

BACKGROUND: Physical restraint rates can be reduced safely in long term care settings, but the strategies used to prevent wandering, falls, and patient aggression have not been tested for their effectiveness in preventing therapy disruption. A restraint reduction program (RRP) consisting of four core components (administrative, educational, consultative, and feedback) was implemented in 1998-1999 in 14 units at two acute care hospitals in geographically distant cities. METHODS: The RRP was targeted at units with prevalence rates of > or = 4% for non-intensive care units (non-ICUs) and > or = 25% for ICUs, as well as two additional units. The RRP was implemented by an interdisciplinary team consisting of geriatricians and nurse specialists. RESULTS: Of the 16,605 admissions to the RRP units, 2,772 cases received RRP consultations. Only six units (four of seven general units and two of six ICUs) demonstrated a relative reduction of > or = 20% in the physical restraint use rate. No increase in secondary outcomes of patient falls and therapy disruptions (patient-initiated discontinuation or dislodgment of therapeutic devices) occurred, injury rates were low, and no deaths occurred as a direct result of either a fall or therapy disruption event. DISCUSSION: Given the minimal success in the ICU settings, further studies are needed to determine effective nonrestraint strategies for critical care patients. ICU clinicians need to be persuaded of the favorable risk-to-benefit ratio of alternatives to physical restraint before they will change their practice patterns. SUMMARY: Efforts to identify more effective interventions that match patient needs and to identify non-clinician factors that affect physical restraint use are needed.     

Incidence and costs of bicycle-related traumatic brain injuries in the Netherlands

The main cause of death and serious disability in bicycle accidents is traumatic brain injury (TBI). The aim of this population-based study was to assess the incidence and costs of bicycle-related TBI across various age groups, and in comparison to all bicycle-related injuries, to identify main risk groups for the development of preventive strategies.Data from the National Injury Surveillance System and National Medical Registration were used for all patients with bicycle-related injuries and TBI who visited a Dutch emergency department (ED) between 1998 and 2012. Demographics and national, weighted estimates of injury mechanism, injury severity and costs were analysed per age group. Direct healthcare costs and indirect costs were determined using the incidence-based Dutch Burden of Injury Model.Between 1998 and 2012, the incidence of ED treatments due to bicycle-related TBI strongly increased with 54%, to 43 per 100,000 persons in 2012. However, the incidence of all bicycle-related injuries remained stable, from 444 in 1998 to 456/100,000 in 2012. Incidence of hospital admission increased in both TBI (92%) and all injuries from cycling (71%). Highest increase in incidence of both ED treatments and hospital admissions was seen in adults aged 55+. The injury rate of TBI per kilometre travelled increased (44%) except in children, but decreased (−4%) for all injuries, showing a strong decrease in children (−36%) but an increase in men aged 25+, and women aged 15+. Total costs of bicycle-related TBI were €74.5 million annually. Although bicycle-related TBI accounted for 9% of the incidence of all ED treatments due to cycling, it accounted for 18% of the total costs due to all bicycle-related injuries (€410.7 million). Children and adolescents (aged 0–24) had highest incidence of ED treatments due to bicycle-related injuries. Men in the working population (aged 15–64) had highest indirect costs following injuries from cycling, including TBI. Older cyclists (aged 55+) were identified as main risk group for TBI, as they had highest ED attendance, injury rate, injury severity, admission to hospital or intensive care unit, and costs.Incidence of ED treatments due to cycling are high and often involve TBI, imposing a high burden on individuals and society. Older cyclists aged 55+ were identified as main risk group for TBI to be targeted in preventive strategies, due to their high risk for (serious) injuries and ever-increasing share of ED visits and hospital admissions.     

Exploitation versus Exploration in Market Competition

A general simulation model of market competition is developed to explore the effectiveness of and interactions between different types of product exploration and exploitation strategies, that is, innovation, imitation and process improvement. The model, like real markets, is highly non-linear such that analytical solutions are not possible. We use simulation experiments to examine firm survival and the effectiveness of different strategies under different market conditions including competitors' strategies, how long it takes for each strategy to bear fruit and how costly it is, and the timing, growth and duration of product life cycles. The model is implemented on the Internet and provides the basis for further experiments to examine the impact of different combinations of firm strategies on survival and performance.     

The advanced practice nurse–nephrologist care model: Effect on patient outcomes and hemodialysis unit team satisfaction

The tertiary care nurse practitioner/clinical nurse specialist (NP/CNS) is an advanced practice nurse with a relatively new role within the health‐care system. It is stated that care provided by the NP/CNS is cost‐effective and of high quality but little research exists to document these outcomes in an acute‐care setting. The clinical coverage pattern by nephrologists and NP/CNS of a hemodialysis unit in a large academic center allowed such a study. Two NP/CNS plus a nephrologist followed two of three hemodialysis treatment shifts per day; only a nephrologist followed the third shift. The influence of this care pattern of patients was examined using a cross‐sectional review of outcomes such as adequacy of delivered dialysis, anemia management, phosphate control, hospitalizations, etc. In addition, the level of satisfaction of the dialysis team and perceptions of care delivered with the care models was assessed. The care model staff‐to‐patient‐number ratio was similar in both groups (1:27 for NP/CNS plus nephrologist; 1:29 for nephrologist alone). Patient demographics were similar in both groups but the NP/CNS–nephrologist group had patients with more comorbidities. No statistically significant (p < 0.05) differences existed between the groups in patient laboratory data, adherence to standards, medications, inter‐ and intradialytic blood pressure, achievement of target postdialysis weights, and hospitalizations or emergency room visits. Significantly more adjustments were made to target weights and medications and more investigations were ordered by the NP/CNS–nephrologist team. Team satisfaction and perceptions of care delivery were higher with the NP/CNS–nephrologist model. It is concluded that the NP/CNS–nephrologist care model may increase the efficiency of the care provided by nephrologists to chronic hemodialysis patients. The model may also be a solution to the problem of providing nephrologic care to an ever‐growing hemodialysis population.     

Measuring transport injuries in a developing country: an application of the capture-recapture method

The purposes of this study are to provide an estimation of the incidence of transport injuries in a defined local community in Nicaragua by using the capture-recapture method, and to compare results using this method when data at different levels of severity are utilized.Two sources of injury data were used to monitor injuries: hospital data (inpatient and outpatient) and traffic police records. Characteristics available for matching included name, age, sex, and date of occurrence. The methodology of capture-recapture was used to estimate the ascertainment degree of both sources of data and the estimate mortality and incidence rates. Estimates were calculated both when all hospital data were taken into account (inpatient and outpatient combined) and when only inpatient records were matched against police records.First, including police records and all hospital data, the mortality and morbidity estimates were 35.5/100000 and 43.7/1000 per year, respectively. Second, when outpatients were excluded from the analysis, the corresponding estimates were 28.6/100000 and 7.5/1000, respectively. In non-fatal cases, the ascertainment-corrected coverage through police records was 2.6% and through hospital surveillance 19.0% when both inpatients and outpatients were included. In fatal cases, the corresponding rates were 56.1 and 22.8%, respectively. The combined data set pointed out pedestrians and cyclists as the main risk groups. Most fatalities were due to head injuries.Our results show that neither police records nor hospital records nor the aggregate database provided acceptable coverage of transport-related injuries. Combining police and hospital data by means of capture-recapture analysis produces more valid estimates, but caution must be given to the issue of severity heterogeneity between the two sources.     

The Patient Care Notebook: quality improvement on a rehabilitation unit

BACKGROUND: Shortened lengths of stay in acute and rehabilitation hospitals, continuing financial pressures on all postacute care services, and increasing out-of-pocket health care costs for patients and families challenge rehabilitation hospitals' patient education and discharge planning processes. Spaulding Rehabilitation Hospital (Boston) introduced a patient care notebook in a 15-bed satellite unit and pilot tested its contribution to the patient education and discharge planning process. DEVELOPING THE NOTEBOOK: The three-ring binder notebook included sections on medical appointments and phone numbers, understanding illness and medical care, coping with illness, physical activities, recommendations for the home, and community resources, with both standard and patient-specific information. RESULTS: Most of the patients and caregivers who received the notebooks found them to be helpful, and most staff indicated that the notebook improved the teaching process. Telephone calls to the unit after home discharges decreased form 28 calls for 11 discharges to 6 calls for 21 discharges after the notebook began to be used regularly. DISCUSSION: Staff felt that the process of using the notebook helped focus attention on teaching during the entire course of a patient's hospitalization rather than just a day or two before discharge. The patient care notebook process is being introduced to the entire hospital and to all patients, regardless of discharge location and the patient's literacy or proficiency with English. CONCLUSION: In using the notebook, the QI team, and the entire unit staff, learned about the complexities of QI, patient education, and discharge planning. The notebook process was implemented throughout the hospital a little more than a year after the completion of the pilot project.     

Overcoming turf battles: developing a pragmatic, collaborative model to improve glycemic control in patients with diabetes

BACKGROUND: Collaboration between primary care physicians (PCPs) and endocrinologists should be the first step in improving care of patients with diabetes. However, the coordination of care between specialists and PCPs often does not work well. At Vanderbilt University Medical Center, a collaborative model between PCPs and endocrinology was used in an effort to improve glycemic control for patients with diabetes. METHODS: In 1998 a project team was formed; the team members attempted to find ways to improve the care of patients with diabetes, specifically patients with poor glycemic control. The team proceeded through ten iterations of the model before reaching one accepted by all-one with clear responsibilities and referral criteria. RESULTS: Survey results indicated a high level of satisfaction with the collaborative model among patients and PCPs. Appropriate referrals to the diabetes improvement program--a 12-week outpatient program consisting of instruction and support in diabetes self-management coupled with adjustment of insulin and oral hypoglycemic medications-increased during the team effort, and a control chart indicated a change in the process that was significant and sustained. The patients enrolled in the program experienced a reduction of mean glycated hemoglobin levels from 9.2% at entry to 7.5% after 3 months (p < 0.05). DISCUSSION: An initial first step to improving care is to create an environment of trust and collaboration between the PCPs and specialists who assist in that care. After this collaboration has been established, many of the improvements identified in other studies can more easily be implemented.     

Dynamic and static job allocation for multi-server systems

We consider the optimal assignment of groups of jobs to a fixed number of time periods over a finite horizon to minimize the total facility idling and job waiting costs. The capacity of the facility varies randomly in the sense that the time that each one of the multiple servers becomes available is random (servers arrive late). The service times are also random and are independent and identically distributed. With approximations, we formulate a dynamic optimization model for this problem. With a simple modification, we can apply this dynamic model to a static outpatient appointment problem. We propose two methods to compute the capacity distribution: (1) Poisson approximation and (2) simulation. While the Poisson approximation works well for exponential service times, the simulation scheme enables us to use the dynamic model without actually specifying the service time distribution. The performance measures of the schedules obtained with these two methods compare well with those of the optimal allocation obtained from (exhaustive) simulation. We also conduct numerical studies to investigate the dynamics between the idling and waiting costs ratio and the number of scheduling periods.     

A case management model to improve hemodialysis outpatient outcomes

Optimal outpatient dialysis care is often difficult to achieve and a case management model to augment conventionally applied nursing and physician resources focusing on continuous quality improvement presents a possible solution to improving outcomes in this setting. We applied this model to patients followed by our physician group. Continuous quality improvement data generated from the dialysis unit database were used to analyze outcomes in patients enrolled in this model. Data from the cohort of patients followed in 2003 served as the reference source for comparative purposes. The nurse case manager assumed responsibility during the second quarter of 2004. Comparing outcomes data from 2005 with data from 2003, we were able to achieve a 3.12% improvement in the annualized mean percent crude mortality per 100 patient years (p<0.003). There was a 3.46-day trend to improvement in patient hospital days per year (p<0.06). The percentage of catheters used as primary access decreased by 9.59% (p<0.025), and the percentage of patients meeting an eKdrt/V goal > or =1.2 increased by 15.33% to 92.37% (p<0.001). These data appear to support the utility of a case manager model in our system.     

基于三维头部数值模型的颅脑碰撞损伤机理研究

头部碰撞载荷会致使颅脑发生创伤性脑损伤（Traumatic Brain Injury,TBI）。其中,脑组织挫裂伤是最为常见的一种,具有高死亡率与高致残率的特性。该文基于数值模拟方法对其开展相关研究,揭示其损伤机理,对该类损伤的预防救治与相关防护设备的开发都具有重要意义。首先,该文基于颅脑的核磁共振切片建立了人体头部三维数值模型,该模型真实地反映了颅脑的生理特征与细节构造。在该模型中,颅骨采用典型类三明治结构进行表征,其内外层为刚度与密度较大的骨密质,中间层为骨松质。为了真实反映脑组织与颅骨间的相互作用,将脑脊液与蛛网膜小梁简化为均质整体,采用状态方程表征脑脊液的液态特性,并通过较小的剪切模量表征蛛网膜小梁的剪切传递作用。然后,基于死尸前额碰撞实验对三维头部数值模型的有效性进行验证。该头部模型采用三种不同的颈部约束边界条件对前额碰撞实验进行数值模拟,模拟结果表明:自由边界条件下的模拟结果与实验数据吻合良好,验证了该头部碰撞模型的有效性;而在竖向约束边界条件或固定边界条件下颈部的约束过于刚硬,导致撞击处与对撞处的颅内正、负压力交替变换,与实验结果相比出现较大偏差。最后,利用验证的头部碰撞模型对枕部碰撞过程进行数值模拟,并结合前额碰撞的模拟结果,分别从脑组织压力（体积变形）与Mises应力（剪切变形）等方面对颅脑的动态响应规律进行分析;进一步结合医学上颅脑碰撞损伤的统计数据,揭示了脑组织挫裂伤的损伤机理,建立了相应的损伤准则。     

Applying the domain management model in treating patients with chronic diseases

BACKGROUND: In the health care system in the United States, the management of chronic health conditions and their functional consequences challenge and frustrate patients, caregivers/families, health care providers, and physicians. Contributing factors include a lack of physician and health care provider training and a health system that emphasizes diagnosis and management of acute illnesses. A broader patient care model is required for patients with chronic disease(s). USING THE DOMAIN MANAGEMENT MODEL (DMM) TO CLASSIFY PATIENTS' CLINICAL PROBLEMS: The DMM is a synthesis of approaches used in internal medicine, geriatric medicine, and physical medicine and rehabilitation. All clinical problems, their treatments, and their outcomes can be classified and followed over time in a multiaxial model with four domains-medical/surgical issues, mental status/emotions/coping, physical function, and living environment. APPLICATIONS OF THE DMM IN MEDICAL RECORD TEMPLATES: Use of the four domain headings in standard templates can lead to an improved awareness of all the relevant issues in the management of chronic illnesses. This awareness precedes a physician's implementation of better care processes. Also, good patient care decisions require good information. MANAGEMENT OF FUNCTIONAL PROBLEMS: The DMM can be used to educate care providers and organize care in terms of important and common functional problem (for example, trouble walking, which lacks a standard approach in health care). CONCLUSION: This common framework for the organization, documentation, and communication of patients' care over time will help teach systematic mangement of chronic health conditions and help with future research on complex patient management.     

Patient with a total artificial heart maintained on outpatient dialysis while listed for combined organ transplant,a single center experience

Advanced mechanical circulatory support is increasingly being used with more sophisticated devices that can deliver pulsatile rather than continuous flow. These devices are more portable as well, allowing patients to await cardiac transplantation in an outpatient setting. It is known that patients with renal failure are at increased risk for developing worsening acute kidney injury during implantation of a ventricular assist device (VAD) or more advanced modalities like a total artificial heart (TAH). Dealing with patients who have an implanted TAH who develop renal failure has been a challenge with the majority of such patients having to await a combined cardiac and renal transplant prior to transition to outpatient care. Protocols do exist for VAD implanted patients to be transitioned to outpatient dialysis care, but there are no reported cases of TAH patients with end stage renal disease (ESRD) being successfully transitioned to outpatient dialysis care. In this report, we identify a patient with a TAH and ESRD transitioned successfully to outpatient hemodialysis and maintained for more than 2 years, though he did not survive to transplant. It is hoped that this report will raise awareness of this possibility, and assist in the development of protocols for similar patients to be successfully transitioned to outpatient dialysis care.     

Understanding patient willingness to recommend and return: a strategy for prioritizing improvement opportunities.

BACKGROUND: Beginning in April 1995, an ongoing, comprehensive measurement system has been developed and refined at BJC Health System, a regional integrated delivery and financing system serving the St Louis metropolitan area, mid-Missouri, and Southern Illinois, to assess patient satisfaction with inpatient treatment, outpatient treatment, outpatient surgery, and emergency care. This system has provided the mechanism for identifying opportunities, setting priorities, and monitoring the impact of improvement initiatives. METHODS: Satisfaction with key components of the care process among 23,361 patients (7,083 inpatients, 8,885 patients undergoing outpatient tests/procedures, 5,356 patients undergoing outpatient surgery, and 2,037 patients receiving emergency care) at 15 BJC Health System facilities was assessed through weekly surveys administered in April 1995 through December 1996. RESULTS: Structural equation models were developed to identify the key predictors of patient advocation-willingness to return for or recommend care. Across all venues of care the compassion provided to patients had the strongest relationship to patient advocation. Within each venue of care, however, a slightly different set of secondary factors emerged. The resulting models provided important information to help prioritize competing improvement opportunities in BJC Health System. In one hospital, a general medicine unit working for several years with little success to improve its patient satisfaction decided to focus on two primary factors predicting patient advocation: nursing care delivery and compassionate care. Root cause analysis was used to determine why two items-staff willingness to help with questions/concerns and clear explanation about tests and procedures-were rated low. On the basis of feedback from phone interviews with discharged patients, the care delivery process was changed to encourage patients to ask questions. Across the next two quarters, this unit experienced significant improvements in both targeted items. DISCUSSION: The significance of compassionate care and care delivery again speaks not only to the importance of the technical quality of clinical care but also to the customer-focused way in which this care was provided. After the primary predictors of patient advocation were identified, management was able to strategically focus improvement initiatives to maximize their impact. Across the organization, improvement teams scanned their data to find key factors where performance was lacking. Once these key opportunities were identified, the teams developed potential solutions and launched initiatives to improve their performance. SUMMARY AND CONCLUSIONS: Results suggest that some core issues are of extreme importance to patients regardless of whether they are receiving care in an inpatient, outpatient, or emergency setting. The compassion with which care is provided appears to be the most important factor in influencing patient intentions to recommend/return, regardless of the setting in which care is provided.     

Handling dataset dependence with model ensembles for skin lesion classification from dermoscopic and clinical images

Dataset dependence affects many real-life applications of machine learning: the performance of a model trained on a dataset is significantly worse on samples from another dataset than on new, unseen samples from the original one. This issue is particularly acute for small and somewhat specific databases in medical applications; the automated recognition of melanoma from skin lesion images is a prime example. We document dataset dependence in dermoscopic skin lesion image classification using three publicly available medium size datasets. Standard machine learning techniques aimed at improving the predictive power of a model might enhance performance slightly, but the gain is small, the dataset dependence is not reduced, and the best combination depends on model details. We demonstrate that simple differences in image statistics account for only 5% of the dataset dependence. We suggest a solution with two essential ingredients: using an ensemble of heterogeneous models, and training on a heterogeneous dataset. Our ensemble consists of 29 convolutional networks, some of which are trained on features considered important by dermatologists; the networks' output is fused by a trained committee machine. The combined International Skin Imaging Collaboration dataset is suitable for training, as it is multi-source, produced by a collaboration of a number of clinics over the world. Building on the strengths of the ensemble, it is applied to a related problem as well: recognizing melanoma based on clinical (non-dermoscopic) images. This is a harder problem as both the image quality is lower than those of the dermoscopic ones and the available public datasets are smaller and scarcer. We explored various training strategies and showed that 79% balanced accuracy can be achieved for binary classification averaged over three clinical datasets.     

Mortality and injury patterns associated with roof crush in rollover crashes

Background

In the United States, a significant number of spine injuries, traumatic brain injuries (TBI), and deaths result from motor vehicle rollover crashes each year though they make up a small percentage of total crashes. We sought to explore the relationship between these injuries and the degree of roof crush.

Methods

We searched the NASS CDS database for belted, adult (≥16), non-middle seat passengers involved in rollover crashes from 1993 to 2006. We also searched the CIREN database for illustrative cases. Logistic regression was used to evaluate the relationship between different levels of roof crush and mortality, severe injury (AIS ≥3) to the spine, spinal cord, and head injury.

Results

The risk of mortality, TBI, and spine injury all increased as the degree of roof crush increased. For mortality increased risk occurred at >15 cm [15-30 cm: OR 2.089 (95% CI: 1.461-2.987); >30 cm: OR 6.301 (95% CI: 4.369-9.087)]. For TBI, increased risk was seen above 15 cm crush [15-30 cm: OR 1.52 (95% CI: 1.045-2.21); >30 cm: OR 3.672 (95% CI: 2.456-5.490)]. For spine injury increased risk was seen above 8 cm crush [8-15 cm: OR 1.968 (95% CI 1.273-3.043); 15-30 cm: OR 2.530 (95% CI 1.634-3.917); ≥30 cm OR 2.682 (95% CI 1.474, 4.877). Results were similar across the different statistical models.

Conclusion

There is an association between the degree of roof crush and mortality, spine injury, and head injury in rollover crashes.