Survival times and case fatality rates of brain-injured persons

Transcription

1 J Neurosurg 63:537-53, 1985 Survival times and case fatality rates of brain-injured persons JESS KRAUS, M.P.H., PH.D., CAROL CONROY, M.P.H., PAMELA COX, M.P.H., KAREN RAMSTEIN, M.P.H., AND DANIEL FIFE, M.D. Division of Epidemiology, School of Public Health, University of California, Los Angeles, California, and Insurance Institute for Highway Safety, Washington, D.C. t,- Survival time after injury (the time from injury to death) imposes an important constraint on the timing of the delivery of postinjury medical care. From a population-based study of brain-injured people, the survival times in 52 cases with fatal outcomes were studied. Prehospital deaths as well as hospital deaths were included. Survival times were considerably shorter for 95 people with untreatable injuries (Abbreviated Injury Scale level 6) than for the remaining 7 whose injuries were potentially treatable. For the former group, the median survival time was l minutes; for the latter, it was 2 hours. For those with potentially treatable injuries, the median time from injury to receiving medical assistance was approximately 3 minutes and 82% received medical assistance within 1 hour of injury. Short survival time was associated with prehospital death, young age, high Injury Severity Score, and having a nonbrain injury as the most severe injury. For patients who arrived alive at a hospital, intracranial surgery was associated with increased survival time. KEY WORDS " brain injury 9 epidemiology 9 Injury Severity Score 9 survival time 9 mortality rate T HE time from injury to death for fatally injured persons may be an important constraint on postinjury medical care improvement. For people who will die soon without medical care, intervention must be initiated early. Studies of injuries that eventually prove fatal may suggest improvements in medical care or may improve understanding of the limits of such care. Such studies may focus on the immediate cause of death ~2 or on the opportunities for alternative courses of medical intervention. 15'~6'25 To accurately assess time to death and case fatality rates, the mortality data of a large group of injured persons are necessary. Excluding those who died at the scene of the injury or those who were pronounced dead on arrival at an emergency room is appropriate only if the subject of interest is in-hospital deaths ~z and not all mortality following injury. The published literature on survival time after injury is sparse, and reports that are available 5'6'12'13'19-21,2 often do not define an early versus a late death, do not state the follow-up period to death or to the end of the study, or do not distinguish between those who were dead on arrival and those who died in the hospital. Such problems complicate comparisons of results from these earlier studies. There have been very few published reports on time to death that focus exclusively on brain-injured people, a group with different diagnostic, treatment, and prognostic features from people with other types of injuries. This report examines case fatality rates and time to death in all brain-injured people in a defined population. Clinical Material and Methods New occurrences of brain injury in 1981 were determined for all usual residents of San Diego County, California, excluding transients but including personnel residing on any of the military reservations. 1 "Brain injury" means physical damage to, or functional impairment of, the cranial contents from acute mechanical energy exchange, exclusive of birth trauma. Persons with fractures of the head or face, or soft-tissue injuries of the head without concurrent brain injury were excluded. To be included, the brain injury had to be diagnosed by a physician either as a result of an autopsy (in the case of immediate or hospital emergency-room deaths and most in-hospital deaths) or following hospital admission. J. Neurosurg. / Volume 63/October,

2 J. Kraus, et al. FIG. 1. Graph showing the interval between injury and death for all fatalities (solid line), for those with untreatable injuries (broken line), and for those with treatable injuries (dotted line). FIG. 2. Graph showing the interval between injury and death for individuals dead on arrival (DOA) at emergency rooms (broken line) and those who died in the hospital (solid line). Brain-injury cases were identified following an examination of the following records: emergency-room, admission, and hospital records of acute-care general hospitals in San Diego County; all coroner's records in San Diego County, and the adjoining counties of Irapedal, Orange, and Riverside; all death certificates (regardless of cause or place of death) for residents of San Diego County; nursing home and extended-care facility records in San Diego County; and the medical records of the nine major hospitals in the three counties bordering San Diego County. Of the 28 acute-care hospitals in San Diego County, 25 hospitals (containing 95.3% of the total number of beds) allowed access to their medical records. No adjustment was made for the cases at the three nonparticipating hospitals. In this study, injuries were described using the 198 revision of the Abbreviated Injury Scale (AIS) which assigns a severity level ranging from 1 to 6 for each injury? Injuries with AIS severity level l (AIS-1) are mild injuries, often not requiring medical attention, such as a closed fracture of a finger or a slight concussion. Those with AIS-3 are serious but treatable, such as a contusion of the spleen or cerebellum. Injuries of AIS-5 are the most severe lesions consistent with survival, such as a lacerated aorta or cerebellum. Injuries with AIS-6 are considered untreatable, such as a laceration or a crushing injury to the brain stem. All persons in the study group had at least an AIS-2 brain injury. The Injury Severity Score (ISS), which is based on the AIS scores, was used to assign an overall injury rating to each person. 3, The ISS classifies injuries into six body regions: face, head and neck, chest, abdomen, extremities, and external. The three regions with the most severe injuries (that is, those with the highest AIS scores) are identified, and the ISS is calculated as the sum of the squares of these three AIS values. The ISS is predictive for the risk of death, and its predictive power is improved by age adjustment. The ISS for cases with an AIS-6 injury is not calculated because, by definition, such injuries are not currently treatable. In this study, the ISS and age were used as standardizing factors in assessing case fatality rates and survival times. Cases were grouped into three severity categories on the basis of their ISS: to 2, 25 to 9, and 5 to 75. Cases were grouped by years into the following age categories: under 15 years, 15 to 3, 35 to 6, and 65 and over. Case fatality rates were standardized by weighting these rates for each age and specific ISS category by the proportion of the total brain-injured population in that category. For patients admitted to the hospital, death or survival was defined by the patient's status at the time of hospital discharge. For all cases, death or hospital discharge occurred within 16 weeks of injury. For patients who died, survival time is defined as the length of time from the date, hour, and minute of injury to the date, hour, and minute of death. Information on time of injury and death was obtained from police, emergency transport, hospital, and/or coroner's records. For purposes of this report, early deaths were defined as those occurring within 2 hours after the time of injury; all other deaths were defined as late deaths. This was consistent with the definition used by Lokkeberg and Grimes.15 First emergency medical contact was defined as the arrival of the patient at the emergency room unless a paramedical ambulance was used, in which case the first medical contact was taken to be the arrival of the paramedics. A multivariate technique I was used to identify the factors that differentiated patients who died following hospital admission into early and late deaths. Stepwise logistic regression was performed using the 1983 Biomedical Data Processing Logistic Regression procedure. 9 Hosmer's goodness-of-fit chi-square was used 538 J. Neurosurg. / Volume 63 / October, 1985

3 Survival times and fatality rates in brain injury TABLE 1 Time from injury to first medical contact for in-hospital deaths* Lapsed Time (min) Cumulative Percent of Deaths 1 3 2O * Those with Abbreviated Injury Scale (AIS) severity level 6 injuries were excluded. Data for San Diego County, California, in FIG. 3. Graph showing the interval between injury and death correlated with the Injury Severity Score (ISS) adjusted for age. to measure how well the data fit the logistic function. The variables used in this analysis were: gender, ISS, age, highest AIS brain-injury severity level, external cause of injury, time from injury to treatment, Glasgow Coma Scale (GCS) score, method of emergency transport, and cranial surgical procedure. Throughout this paper the expression "immediate cause of death" is used to indicate the proximate (as opposed to underlying) cause of death. Immediacy in the sense of time is not intended by this expression. The immediate cause of death for persons without AIS- 6 injuries was coded according to the International Classification for Disease (ICD9-CM), using information in the coroner's report or the death certificate. For purposes of this study, six patients who died of cerebral edema (five within 2 hours of injury) were classified as deaths from brain injury. Results During 1981, there were 3358 new cases of brain injury (including 562 cases with fatal outcome) among San Diego County residents; 261 (6%) were dead at the scene of the injury and were transported directly to the morgue, 125 (22%) were dead on arrival at the emergency facility, and 176 (31%) died in a hospital. Twenty people (%) who died in isolated places and were not found until days or weeks later were excluded from the present study because of uncertainty about the time of injury or death. Hence, the present study is based on 3338 brain-injury cases, 52 of which were fatal. The overall mortality rate was 16.2%; most deaths occurred soon after injury. Nearly 7% of all braininjured people (1% of those who were fatally injured) died within 1 minutes of injury. The death rate decreased progressively with the period of survival time after injury, a decrease that was relatively smooth after the first l minutes postinjury (Fig. l). The survival curve for the 95 people with an untreatable (AIS-6) injury differed markedly from that for the 7 with treatable injuries. Among those with untreatable injuries, 65% had died by 1 minutes postinjury and 95% by 2 hours postinjury. Among those with treatable injuries, 35% of the deaths had occurred by l minutes postinjury and 55% by 2 hours postinjury. Thus, people with untreatably severe injuries had a markedly different distribution of survival time from those who died with treatable injuries. Because this study was concerned with potentially treatable cases, those with AIS- 6 injuries were not included in the remaining analyses. Among fatalities, the place of death (at the scene, during transport, or at a hospital) depended on the survival time. Of those who died in the 1st hour after injury, 8% died at the scene of injury and were transported directly to the morgue, 2% were dead on arrival at an emergency department, and none died after admission to a hospital. Of those who died in the 2nd hour after injury, 2% died at the scene of the injury and were transported directly to the morgue, 9% were dead on arrival at an emergency department, and only 8% died after admission to a hospital. Of those who survived 2 to hours after injury, 6% died at the scene of injury and were transported directly to the morgue, 56% were dead on arrival at an emergency department, and 38% died after admission to a hospital. The distributions of times from injury to death are shown in Fig. 2 for those who were dead on arrival at an emergency department and for those who died after admission to a hospital. Because very few in-hospital deaths were observed for the first 2 hours after injury, the prehospital history of fatal cases was examined in more detail. The time from injury to first medical contact is shown in Table 1. The median time to contact is under 2 minutes; 75% were in contact with medical aid within the first minutes, and 82% within the 1st hour postinjury. Survival time is related to both injury severity and age. After adjustment for age, survival times were shorter for those with high ISS's than for those with low scores (Fig. 3). Individuals with ISS values of to 2 had longer survival times than those with higher scores. The mortality curves for the group with ISS values of 25 to 9 and the group with ISS values of 5 to 75 were J. Neurosurg. / Volume 63 / October,

4 J. Kraus, et al. TABLE 2 Immediate causes of death by survival time* FIG.. Graph showing the interval between injury and death correlated with the individual's age adjusted for Injury Severity Score. Those with Abbreviated Injury Scale severity level 6 (AIS-6) are excluded. similar for the first 2 hours, but thereafter the group with the lower score had longer survival times. Agespecific survival times, adjusted for severity of injury (ISS), are given in Fig.. The median survival time first decreased with age and then increased. For those under 15 years of age the median survival time was 12 hours, for those aged 15 to years it was 1 hour, for those aged 5 to 6 years it was 1 minutes, and for those aged 65 years and older it was hours. Although each individual in our study group had a brain injury, the immediate cause of death may have been an injury to some other body region. The cases were therefore classified according to the immediate cause of death (Table 2). Those whose immediate cause of death was injury to some organ other than the brain tended to have shorter survival times than those whose immediate cause of death was a brain injury. The longest survival times were found among those who died from complications of trauma. As an alternative to examining the immediate cause of death, cases were classified according to the site of the most severe injury (brain, some other body region, or both: that is, deaths equally attributable to injuries of the brain and to some other body region). The results were similar to those stated previously. Patients whose most severe injury was to the brain tended to have longer survival times than other patients (Fig. 5). All persons who died with an ISS of to 2 had an injury of that level either to the brain or to the brain plus another region (Fig. 5 upper). Of 77 individuals who died with an ISS of to 2, 9 (6%) died in the hospital. In these patients, respiratory failure occurring more than 2 hours after injury was the cause of 18 deaths; nine of these were among patients less than 65 years old. For those with an ISS of 25 to 9 (Fig. 5 center), individuals with the most severe injury to the brain had longer survival times than those with the Immediate Cause ICD Postinjury Survival Time of Death No. <8 Hrs 8-2 Hrs >2 Hrs Total brain injury multiple skull fracture cerebral laceration/ contusion brain hemorrhage other intracranial injury cerebral edema brain trauma other organ trauma multiple organ (extreme) hemothorax abdominal organ cervical spinal cord early complications other organ failure/systemic respiratory failure bronchopneumonia heart failure cardiac dysrhythmia sepsis peritonitis all causes * Data are for brain-injured persons with fatal outcomes recorded in San Diego County, California, in ICD = International Classification for Disease. most severe injury to some other body region. Those with most severe injuries to both regions had survival times similar to the former group. For patients with an ISS between 5 and 75 (Fig. 5 lower), those whose most severe injury was to the brain had longer survival times than those whose most severe injury was to a nonbrain region. Those with most severe injuries to both regions had survival times similar to the latter group. Although the survival times are presented in ISS specific score ranges ( to 2, 25 to 9, and 5 to 75), it is possible that within each range there is considerable variation between the ISS values of the groups being compared. The within-range mean ISS for persons whose most severe injury was to the brain (including those with equally severe injuries to brain and nonbrain regions) was 1, 28, and 59 for the three specific ISS ranges. The within-range mean ISS for the remaining patients (those whose most severe injury was not to the brain or was equal for brain and nonbrain regions) was 12, 3, and 56, respectively. There were no significant differences between mean ISS values within each range. The finding that no patients with an ISS value of to 2 died unless a brain injury was one of their most severe injuries raised the question of whether the ISS prognosis depends on the regions injured. As shown in Fig. 5, within each ISS group, after age adjustment the 5 J. Neurosurg. / Volume 63/October, 1985

5 Survival times and fatality rates in brain injury worst prognosis was faced by patients whose most severe injury included the brain. One of the fairly common interventions used in treating brain-injured patients is cranial surgery. Although such surgery may be intended to be curative, it may also be used in an effort to gain time during which a dangerous intracranial process may resolve. If such surgery is effective in gaining time, it should extend the survival time of the patients who eventually died. Among patients who reached the hospital alive, the best discriminators of early versus late death, based on stepwise logistic regression, were ISS, GCS score, and cranial surgery. The odds ratio of a late death resulting in those who had cranial surgery, adjusting for the linear effects of GCS score and ISS, was 1.9 (95 % confidence interval of 1.2, 3.). Hosmer's goodness-of-fit chi-square for the logistic model was 7.91, df = 8, p value =. (a large p value indicates a good fit). Thus, cranial surgery recipients were almost twice as likely to die later. However, this finding should be interpreted with caution because some very early hospital deaths may have occurred before cranial surgery could be initiated. Discussion Studies of survival time postinjury 2,s,7,1JAg,2~ have appreciable differences in their selection of cases and study methodology. Despite these differences, one major finding is consistent: a substantial proportion of severely injured people die at the scene of the injury or before arriving at an emergency facility. Trunkey and Lim 2 reported that 6% of trauma victims in San Francisco in 1972 were dead at the scene of injury and were admitted directly to the morgue, and Frey, et al., ~ reporting a series of trauma victims from a county in Michigan, stated that 7% were dead at the scene of the injury. Baker and associates 2 reported that of individuals who died from general trauma in San Francisco in 1977, 53% died at the scene of the injury. In the present study, % of the patients with treatable injuries died at the scene of the trauma and were transported directly to the morgue. An additional 23% were dead on arrival at the emergency department. Almost 38% of the deaths occurred after hospital admission. Moreover, nearly all deaths in the first 2 hours postinjury were pre-hospital deaths. This relatively long interval appeared to be due to two causes. First, individuals arriving alive at the hospital may have received supportive and resuscitative efforts that caused them to survive at least 2 hours postinjury. Second, approximately 2% of the patients who died in the hospital reached medical care more than 1 hour postinjury, possibly because of a delay in summoning help. Many of these people were injured in remote areas or were elderly people injured in fails. The problem of delayed medical attention deserves further study. In any such study, it would be important to distinguish between a delay in summoning medical help and a delay in its arrival once summoned. FIG. 5. Graphs showing the interval between injury and death correlated with the region of injury that was the immediate cause of death and the Injury Severity Score (ISS). In this context, a broader problem arises. It is widely accepted that prompt treatment of injuries is important, and military experience with battlefield injuries appears to support this. However, the length of time from injury to treatment does not appear to be indicative of a fatal outcome among brain-injured civilian patients. 15 In view of this finding, it may be worthwhile to better document the subgroups of patients and injuries for whom the time to treatment is important to survival. J. Neurosurg. / Volume 63~October,

6 J. Kraus, et al. The current AIS-ISS system, originally developed to assess trauma from motor-vehicle crashes, is a useful predictor of outcome and of survival time, but, even after adjustment for age, the mortality risk and survival time at a given ISS value depend on the body region with the most severe injury. In particular, people whose most severe injury is a brain injury are more likely to die from their trauma and at a later time than do other people with the same ISS value. Other authors 17'22 have also observed that brain-injured people have higher mortality rates than do those without brain injuries. The AIS may need further revision to adjust for the different prognosis of similarly graded injuries to the brain and to other regions. The possibility that these differences may arise from difficulties in assigning AIS levels to injuries is also worthy of examination. The findings from this current report confirm the conclusions of earlier studies 6'7'1~ that those with the most severe injuries, as judged by the ISS, die sooner than those with less severe injuries. These findings also confirm earlier findings 6'7'1~ that older persons live longer after the injury before they die. It has been suggested that older persons die relatively late because they die of relatively mild injuries. However, even after survival times are adjusted for the ISS, we found that older adults died later than did young adults. Thus, the difference in survival time does not appear to be entirely due to differences in severity of injury among the fatalities. Age-dependent differences in the incidence and outcome of complications of injury may provide an explanation of the age-dependent differences in mortality and survival times, z~ Analysis of immediate causes of death in this current report yields results generally similar to those of earlier studies.2,11.12,2,22,23 A substantial proportion of early hospital deaths are caused by acute organ damage. Most of these injuries are extensive, and persons with multiple trauma may not survive regardless of age or optimum medical care. However, about 15% of all fatally injured persons died of causes generally described as "complications" and most of these deaths occurred more than 2 hours postinjury. It is not clear why patients with treatable injuries die from immediate causes (such as pneumonia, sepsis, or some organ failures) that, at least on the surface, are medically manageable; this question deserves serious attention. Conclusions In summary, the major findings of this study include: 1. Most brain-injured people with fatal outcomes die soon after injury; 1% die in the first 1 minutes. 2. For brain-injured people with fatal outcomes, the median time from injury to first medical treatment is less than 2 minutes, but 18% first receive medical assistance 1 hour or more after injury. 3. Thirty-one percent of brain-injured people with fatal outcomes reach a hospital alive. Such people usually survive at least 2 hours postinjury.. Among brain-injured people with potentially survivable injuries and fatal outcome, long survival time is associated with relatively mild injury, extremes of age (young or old), and having a brain injury as the most severe injury sustained. 5. Among brain-injured people with potentially survivable injuries, the chance of ultimate survival is decreased if the brain injury is the most severe injury sustained. This remains true even after adjustment for age and ISS. 6. Among brain-injured people who reach a hospital alive but ultimately die, cranial surgery is associated with an increase in survival time. 7. The immediate cause of death depends on the survival time. For those with short survival times, most of the deaths are directly due to injuries. For those with relatively long survival times, injury remains the underlying cause of death but the immediate cause of death is usually respiratory failure or infection. Acknowledgments The authors appreciate the comments by William Haddon, Jr., M.D., President, and Brian O'Neill, Senior Vice President of the Insurance Institute for Highway Safety, made during the initial drafts of this report. Editorial assistance given by Sharon Rasmussen, of the Insurance Institute for Highway Safety, is also appreciated. References 1. Afifi A, Clark V: Computer-Aided Multivariate Analysis. Belmont, Calif." Lifetime Learning Publications, 198, pp Baker CC, Oppenheimer L, Stephens B, et al: Epidemiology of trauma deaths. Am J Surg 1:1-15, Baker SP, O'Neill B: The Injury Severity Score: an update. J Trauma 16: , Baker SP, O'Neill B, Haddon W Jr, et al: The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1: , B~ O: Road Casualties: An Epidemiological Investigation. Oslo: Scandinavian University Books, Bull J: The Injury Severity Score of road traffic casualties in relation to mortality, time of death, hospital treatment time and disability. Accid Anal Prey 7:29-255, Clifton GL, McCormick WF, Grossman RG: Neuropathology of early and late deaths after head injury. Neurosurgery 8:39-31, Committee on Injury Scaling: The Abbreviated Injury Scale, 198 Revision. Morton Grove, Ill: American Association for Automotive Medicine, Dixon W J, Brown MB, Engleman L, et al (eds): BMDP Statistical Software Printing with Additions. Berkeley, Calif." University of California Press, 1983, pp Fife D, Davis J, Tate L, et al: Fatal injuries to bicyclists: the experience of Dade County, Florida. J Trauma 23: , I. Frey CF, Huelke DF, Gikas PW: Resuscitation and survival in motor vehicle accidents. J Trauma 9:292-31, Goris RJA, Draaisma J: Causes of death after blunt trauma. J Trauma 22:11-16, J. Neurosurg. / Volume 63 / October, 1985

7 Survival times and fatality rates in brain injury 13. Hutchinson TP: Factors affecting the times till death of pedestrians killed in road accidents. Injury 6:28-212, Kraus JF, Black MA, Hessol N, et al: The incidence of acute brain injury and serious impairment in a defined population. Am J Epidemiol 119:186-21, Lokkeberg AR, Grimes RM: Assessing the influence of non-treatment variables in a study of outcome from severe head injuries. J Neurosurg 61:25-262, Lowe DK, Gately HL, Goss JR, et al: Patterns of death, complication, and error in the management of motor vehicle accident victims: implications for a regional system of trauma care. J Trauma 23:53-59, Oreskovich MR, Howard JD, Copass MK, et al: Geriatric trauma: injury patterns and outcome. J Trauma 2: , Ottosson A, Krantz P: Traffic fatalities in a system with decentralized trauma care. A study with special reference to potentially salvageable casualties. JAMA 251: , Rish BL, Dillon JD, Weiss GH: Mortality following penetrating craniocerebral injuries. An analysis of the deaths in the Vietnam Head Injury Registry population. J Neurosurg 59:775-78, Robertson JS, Tonge JI: Duration of survival in traffic accident fatalities. Med J Aust 2: , Sevitt S: Fatal road accidents in Birmingham: times to death and their causes. Injury : , Smith HPR: Time to die from injuries received in road traffic accidents. Injury 2:99-12, Trunkey D, Lim R: Analysis of 25 consecutive trauma fatalities: an autopsy study. J Am Coil Emerg Phys 3: , Trunkey D, Siegel J, Baker SP, et al: Panel: current status of trauma severity indices. J Trauma 23:185-21, 1983 (see Fig. 9, p 191) 25. Tsuchihashi M: Epidemiological study of the survival time of persons killed in road traffic accidents. Nihon Univ J Med 25:85-13, Wright CS, McMurtry RY, Pickard J: A postmortem review of trauma mortalities -- a comparative study. J Trauma 2:67-68, 198 Manuscript received December 27, 198. This work was supported by the Insurance Institute for Highway Safety. The opinions, findings, and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of the Insurance Institute for Highway Safety. Address for Dr. Fife: Insurance Institute for Highway Safety, Watergate 6, Suite 3, Washington, D.C Address reprint requests to." Jess F. Kraus, Ph.D., Division of Epidemiology, School of Public Health, University of California, Los Angeles, California 92. J. Neurosurg. / Volume 63/October,