The regionalization or integration of trauma care in a specific geographic region involves the organization of all trauma care services into one comprehensive system. The essential requirements of regionalized trauma care systems are: (1) the classification of all hospitals in the region according to the level of trauma care that they could provide as tertiary (Level I), secondary (Level II), and primary (Level III); (2) the designation of tertiary (Level I) trauma centers as the hospitals where patients with major trauma should be treated; (3) establishment of emergency medical services that provide prehospital care to trauma patients, including on-site assessment and treatment; (4) implementation of triage protocols for the transport of trauma patients to the appropriate trauma hospitals; and (5) centralized coordination and control of prehospital and in-hospital care.
The process of trauma care regionalization is usually gradual, with variation in the level of integration and the order in which the various components are implemented. Despite these variations, the general goal of regionalized trauma care is the establishment of an integrated, coordinated system that would provide the best possible care to trauma patients in the community. To realize this goal, trauma care systems may be designed to address the specific needs of the population. [1-5]
The concept of civilian trauma care regionalization was adopted from experiences in military conflicts and originated during the early 1970s in response to the recognition that injuries were a major and neglected disease of modern society. Several studies conducted primarily in the United States showed that trauma care was deficient and that major changes were required. The Emergency Medical Services Act was a legislative measure aimed at implementing regionalized trauma care systems in the United States. As a result, since the enactment and final amendment of this law approximately 300 regionalized emergency medical systems with regionalized integrated trauma care have been established in the United States. [6-9]
The benefits of trauma care regionalization have been supported by several studies that have suggested associations between the implementation of trauma care systems and a decline in trauma-related mortality. The majority of these studies compare the mortality rates in trauma patients before and after the implementation of trauma care systems. Others draw their conclusions from comparisons of regions with and without trauma care regionalization. [10-23]
The general consensus from the studies evaluating trauma care systems is that regionalization reduces mortality because patients are treated at appropriate hospitals and because the time between injury and definitive in-hospital care is minimized. More specifically, these studies have shown that to reduce trauma-related mortality, patients with major, life-threatening injuries should be treated at tertiary (Level I) trauma centers. The data from these studies have also strongly suggested that for patients with major trauma, definitive in-hospital care should be initiated within 60 minutes.
Although taken collectively, the data from these studies have supported the benefits of trauma care regionalization, the causal relationship has been indirectly demonstrated. This is because the majority of these studies are based on specific groups of patients or focus only on specific hospitals. In addition, the comparisons between different regions, or between different time periods within the same region, are often not adjusted for variations in the patient profile or other factors that may affect the study results. One other issue is that the majority of these studies have been conducted in the United States, thus limiting the generalization to other populations.
Establishment of regionalized trauma care systems in Canada began almost 10 years after their introduction in the United States. The organization of Canadian trauma care systems is, in general, similar to that in the United States. Within Canada, however, there exists significant variation with respect to the level of integration of trauma care services and the nature of trauma care available. The level of trauma care organization in Canada may vary from none, to intermediate (where all or some of the components are in place), to complete integration. [24-26] There are no national guidelines for trauma care systems and no legislation to enforce regionalization.
In Quebec, a Canadian province with a population of approximately 8 million, the process of trauma care regionalization was initiated in 1993. Before that time, trauma care was not organized at any level. In urban areas, patients were transported by public or private ambulances to the hospital that was nearest to the site of the event. Although some hospitals had the expertise and technology to treat patients with major trauma, there was no formal designation of trauma centers.
Data from a study conducted in 1988 identified significant differences in the trauma care in Montreal, the largest city in Quebec, with a population of approximately 3 million. [27,28] Prompted by these results, the Quebec government identified trauma as a priority and in 1993 initiated the process of trauma care regionalization. The first step of this process was the designation of four tertiary (Level I) trauma centers. Three of these are in Montreal and one is in Quebec City, the second largest city in the province. Between 1993 and 1996, there were 33 secondary (Level II) and 30 primary (Level III) trauma centers designated. In 1995, transport protocols were introduced according to which patients with major injuries are either transported directly from the scene or are transferred from other hospitals to one of the tertiary trauma centers. In Quebec, a tertiary trauma center is an acute-care hospital that is primarily dedicated to the treatment of trauma patients. Tertiary trauma centers have 24-hour coverage by emergency physicians, a trauma team, general surgery, neurosurgery, orthopedics, and anesthesiology. All other surgical specialists are available within 30 minutes. These hospitals have operating rooms and diagnostic facilities that are devoted to trauma. In addition, patients with major injuries take priority over all other acute cases. Trauma-related training and research are also essential for tertiary trauma centers. Secondary trauma centers have 24-hour coverage by emergency physicians, with surgical specialists available on an on-call basis. Primary trauma centers have emergency room coverage and minimal coverage by other specialists. Stabilization centers have only rudimentary emergency care, with almost no surgical support available. Patients with major trauma who are transported to primary or stabilization centers should be transported to tertiary or secondary centers for definitive treatment.
With respect to prehospital care, there is significant variation within Quebec. In Montreal, a limited number of physicians are employed by the emergency medical system (Urgences-sante) and are dispatched to the scene for cases of major trauma. Physicians provide on-site care to patients that could range from Basic to Advanced Life Support, according to their judgment and without any standardized protocols. Emergency Medical Technicians (EMTs) are also employed by Urgences-sante and are dispatched to the site for cases of minor injuries and cases of major trauma when a physician is not available. The on-site care provided by EMTs is limited by law to Basic Life Support. There exists in Montreal, therefore, a random variation with respect to the on-site trauma care that is provided to patients. In Quebec City, EMTs provide prehospital on-site care with interventions limited to Basic Life Support procedures only. Regardless of the specific structure and personnel composition of prehospital care, since 1993 changes and policies have been implemented aimed at reducing prehospital time and providing accurate triage of major trauma patients to appropriate trauma centers.
The rationale for the present study is based on the following points. First, to date there has not been a study that followed and evaluated a trauma care system from the initiation of regionalization. Second, the process-outcome association between the changes involved in trauma care regionalization and reduced mortality has not been directly and comprehensively demonstrated using data from the same region. Third, the impact of trauma care regionalization in Canada has not been evaluated.
The goals of the current study were to evaluate the impact and assess the process-outcome association of trauma care regionalization in a Canadian region. The current study followed the evolution of trauma care regionalization in Quebec from 1992, or before its inception in 1993, to 1998, measuring its impact on mortality among patients with major trauma. In addition, the prospective nature of the study and the uninterrupted follow-up during the first 5 years of regionalization and integration of the trauma care system will allow the assessment of the process-outcome associations. This assessment will allow the evaluation of causal relationships between the components of a regionalized trauma care system and trauma-related mortality.
PATIENTS AND METHODS
Study Design
This was a prospective cohort study. Patients entered the cohort at the time of injury and were followed to discharge from the acute-care hospital.
Patients
All patients treated for injuries at acute-care hospitals in Montreal and Quebec City between April 1, 1992, and March 31, 1998, were eligible for inclusion in the study. The target population was that of patients with major injuries. To be included in the final study sample, therefore, the patients had to meet one of the following criteria: (1) death as a result of the injury; (2) Injury Severity Score (ISS) > 12; (3) Pre-Hospital Index > 3; (4) two or more injuries with Abbreviated Injury Scale scores >or=to 3; or (5) hospital stay of more than 3 days.
Patients who died at the scene were excluded from the final study sample, although patients who expired after arrival at the hospital but before any admission or emergency room evaluation were included. The rationale for excluding patients who died on the scene is based on the fact that these are immediate deaths that are nonpreventable. The assumption is that these deaths are not affected by any changes in trauma care. Exclusion of these patients, therefore, will not introduce any bias into the study results.
Outcome Measure
The outcome measure of the study was death as a result of injury, defined as any death occurring during the hospitalization.
Study Hypotheses
There were two hypotheses being tested in this study: (1) that trauma care regionalization is associated with a reduction in trauma-related mortality; and (2) that changes in the components of the Quebec trauma care system (process) are associated with reductions in trauma-related mortality (outcome). The components of the trauma care system that define the process of regionalization are: (i) establishment, designation, and support of tertiary trauma centers; (ii) appropriate triage of major trauma patients to tertiary trauma centers; and (iii) reduced prehospital time or delay to definitive in-hospital care. To test the association between the process of trauma care regionalization and mortality, the following intermediate hypotheses were tested: (a) treatment at tertiary (Level I) trauma centers is associated with a reduction in trauma-related mortality; (b) as the Quebec trauma care system evolves toward complete integration and regionalization, the rate of appropriate patient triage will increase and the total prehospital time will be reduced; and (c) reduced prehospital time or delay to definitive in-hospital care is associated with reduced mortality.
Statistical Methods
Hypothesis 1
This hypothesis tested the association between the evolution toward complete integration and regionalization of the trauma services in Quebec and mortality. The fiscal year (April 1 to March 31) was used as the exposure or independent variable describing the level of integration of the trauma care system. The data in Table 1 describe the process of trauma care regionalization and define its important phases. The information in this Table showsthat trauma care regionalization was initiated after the 1992-1993 fiscal year. This period, therefore, was considered the preimplementation or baseline phase. The first change in the establishment of the new trauma care system was the designation of the four tertiary trauma centers, which occurred in 1993-1994. During 1994-1995, these hospitals progressed toward their evolution as tertiary trauma centers by defining trauma as a priority and devoting their resources and expertise to the treatment of trauma patients. During this period, the secondary trauma centers were identified and patient triage protocols become informally operational. According to these protocols, patients with major injuries were preferably, but not exclusively, transported from the scene to a tertiary trauma center. The period between 1993 and 1995 has been described as phase I, or the initiation phase of trauma care regionalization in Quebec. The introduction of trauma centers and the initiation of a trauma care network characterized this phase. An important change during this period was the introduction of guidelines that were aimed at reducing prehospital times.
;)
Table 1: Evolution of the Quebec Trauma Care System
Between 1995 and 1997, the tertiary trauma centers continued to progress in their evolution by concentrating and dedicating their resources to the treatment of trauma patients. During this period, patient triage and transfer protocols were introduced. According to these protocols, patients with severe injuries should be either transported directly to a tertiary trauma center or transferred to one after being initially transported to less specialized facilities. Specific agreements were established between the tertiary centers and surrounding hospitals that ensured compliance with this transfer protocol. Also during this period, 33 secondary, 30 primary, and 32 stabilization trauma centers were designated. In addition, specialized tertiary centers for the treatment of patients with spinal cord and head injuries were identified. This phase, which was characterized by the continued improvement in the care available at the tertiary centers, the introduction of highly specialized facilities, and the implementation of protocols ensuring that patients are treated at the appropriate trauma centers, was designated phase II, or the intermediate phase of regionalization.
The next phase in the evolution of the Quebec trauma care system was the most recent one and was characterized by a review and evaluation of the system, redesignation of trauma centers, and further implementation of the trauma care network, with integration of all hospitals and emergency medical services. This phase was designated phase III, or the advanced phase of regionalization.
To test the first hypothesis, the following statistical analyses were conducted: (1) a comparison of the mortality rates during each fiscal year of the periods before and after regionalization; and (2) a comparison of the mortality rates during each phase of the implementation of the regionalized Quebec trauma care system. Logistic regression models were used to produce relative risk estimates from the relative odds for mortality for each year after the implementation of regionalization. For these analyses, the 1992-1993 period was used as the baseline or reference group. The statistical significance of the odds ratios was evaluated with 95% confidence intervals. The covariates entered in the logistic regression models were: (1) patient age; (2) mechanism of injury; (3) body regions injured; (4) number of injures; and (5) Injury Severity Score.
Hypothesis 2a
This hypothesis tested the mortality rates of patients treated at tertiary centers compared with patients treated at less specialized hospitals. A second analysis compared the mortality rates of patients who were transported directly from the scene to a tertiary trauma center with patients who were transported initially to a secondary or primary center and subsequently transferred to a tertiary center.
Hypothesis 2b
The outcome measures used in testing this hypothesis were: (1) the proportion of patients with major trauma, defined as ISS >or=to 25, who were treated in tertiary trauma centers; (2) the total prehospital time, defined as the time elapsed between injury and arrival of the patient at the hospital (the time of injury was estimated from the time that the call was received at the emergency medical system); and (3) the time interval between discharge from the emergency department and admission to the ward, the intensive care unit, or surgery.
In testing this hypothesis, differences in these outcome measures between patients treated during each fiscal year and each phase of regionalization, as described in Table 1, were evaluated. The chi squared statistic was used to evaluate difference with respect to the proportion of major trauma patients treated at the tertiary centers. Analysis of variance was used to compare the mean prehospital times and the mean times to admission.
Hypothesis 2c
In testing this hypothesis, logistic regression analysis was used to test the association between total prehospital time and time to admission and the risk of mortality. Because of the large sample size, only differences or associations with an alpha level of 1% or less were considered statistically significant.
Data Collection and Management
The patients included in the final study sample were identified after reviewing the medical records of all acute-care hospitals in Montreal and Quebec City. The Quebec Trauma Registry was also used to identify patients from the tertiary trauma centers. The first step in the selection of the sample was a review of the records of all patients treated for injuries as defined by the International Classification of Diseases, Ninth Revision, E-code or as indicated in the emergency room log. Patients fulfilling the study criteria were entered into the final sample. The data were abstracted on standardized data-collection forms by research assistants or medical archivists who were trained in injury coding and had experience with trauma patient charts. After abstraction, the data were entered in a computer using customized data-management software that was developed for this study. The information on prehospital times was verified when possible from the records of the emergency medical services. The final database was converted to SPSS data sets for analysis.
RESULTS
The study sample was composed of 12,208 patients who met the inclusion criteria. The distribution of patients by study year is shown in Table 2. The data in Table 3 show that the mean age of the patients decreased steadily and significantly, from 54 to 46 years, during the study period. This is also evident by the increase in the proportion of patients younger than 25 years, from 18% to more than 25%, and the decrease in the proportion of patients older than 75 years, from 29 to 16%. The changes in both mean age and age distribution are clinically and statistically significant (p < 0.001). During the 6-year period covered by the study, the male/female ratio of the patients remained constant at 2:1, with approximately two-thirds of the patients being male.
Table 2: Number of patients by study year
Table 3: Demographics data by study year
The mean Injury Severity Score increased from 25.5 in 1992-1993 to 27.5 in 1997-1998. There was a fluctuation in this parameter during the study period, however, which would suggest that this increase does not indicate a true trend despite the statistically significant differences between study years (p = 0.01). A similar observation was made for the change in ISS distribution during the study period (Table 4).
Table 4: Injury Severity Score by study year
With respect to the body regions injured, the proportion of patients with injuries to the head, thorax, or abdomen increased significantly during the study period (p = 0.001). Throughout the study period, the majority of the patients, between 58 and 72%, sustained injuries to the head or the extremities. Thoracic injuries were the second most common, occurring in 36 to 47% of the patients, followed by injuries to the abdomen, which occurred in 23 to 30% of the patients (Table 5).
Table 5: Body regions injured by study year
During the study period, motor vehicle collisions and falls were the mechanisms of injury for more than 80% of the patients. The proportion of patients injured in motor vehicle collisions increased from 44% in 1992-1993 and 41% in 1993-1994 to more than 50% in 1997-1998. This increase is statistically significant (p = 0.001) and also represents an important change in the profile of injuries. Conversely, the proportion of patients injured by falls decreased steadily and significantly during the study period, from 47% in 1992-1993 to 34% in 1997-1998 (p = 0.001). An increase was also observed in the proportion of patients who were injured with blunt objects, but this remained a relatively small group, representing between 2 and 7% of the total sample (Table 6).
Table 6: Mechanism of injury by study year
The primary outcome measure of the study was survival to hospital discharge. The data in Table 7 show that the proportion of patients who were discharged alive increased from 48% in 1992-1993 to 82.3% in 1997-1998. This represents a significant threefold decrease in the mortality rate, from 52% in 1992-1993 to approximately 18% between 1996 and 1998 (p < 0.001). It is important to note that there was a steady decline in the mortality rate for every year of the study since the initiation of regionalization (Figure 1).
Table 7: Discharge status by study year
Figure 1: Mortality by study year.
A similar pattern of decreasing mortality was observed when the data were analyzed according to the phase of trauma care regionalization in Quebec, as defined in Table 1. These results, which are summarized in Table 8, show that the mortality rate decreased from 52% before regionalization, to 32% during the initial phase of regionalization, to 19% during the intermediate phase, and to 18% during the advanced phase. These changes are both clinically and statistically highly significant (p < 0.001).
Table 8: Discharge status by regionalization phase
The data in Table 9 summarize the discharge status of the patients by study year stratified by Injury Severity Score categories. These data show that the mortality rate in patients with moderate injuries (ISS = 12-24) decreased significantly, from 28% in 1992-1993, to 13% in 1993-1994, to 8% in 1994-1995 (p < 0.001). The mortality rate in these patients stabilized at 3% between 1995 and 1998. For patients with major trauma (ISS = 25-49), the mortality rates continued to decline significantly, from 66% in 1992-1993 to 16% in 1997-1998 (p < 0.001). The mortality rate in patients with fatal injuries (ISS = 50-75) remained fairly constant during the study period at approximately 70%. The data in Table 10 show the same pattern of decline in mortality during the different phases of trauma care regionalization.
Table 9: Discharge status by study year, stratified by Injury Severity Score category
Table 10: Discharge status by regionalization phase, stratified by Injury Severity Score category
The data in Table 11 and Figure 2 summarize the prehospital and admission delays during the study period. Prehospital time was defined as the interval from the time that the call was received at the emergency medical system to the time that the patient arrived by ambulance at the first receiving hospital. Time to admission was defined as the interval between the patient's arrival in the emergency room and the time when the patient was discharged from the emergency room to a ward, an intensive care unit, or the operating room. The data show that the mean prehospital time decreased from 62 minutes in 1992-1993 to 44 minutes in 1997-1998. This decrease of 18 minutes is clinically extremely important given that it represents a 30% reduction in prehospital time, and it is also statistically significant (p < 0.001). A similar but less consistent pattern was observed for time to admission, which decreased from between 152 and 167 minutes in the years 1992 to 1994 to between 114 and 128 minutes in the years 1996 to 1998. The data in Table 12 show that the pattern of reduction in prehospital and admission delays was similar when the data were analyzed by phase of regionalization. These data also show a more marked reduction in the time to admission during the intermediate and advanced phases of regionalization. The changes in both prehospital time and time to admission during the regionalization phases were statistically significant (p < 0.001).
Table 11: Prehospital and admission delays by study year
Figure 2: Prehospital and admission delays by study year.
Table 12: Prehospital and admission delays by regionalization phase
The data in Table 13 and Table 14 and Figure 3 describe the Injury Severity Score distribution of the patients treated at the tertiary (Level I), secondary (Level II), and primary (Level III) trauma centers during the study period. These data show that the proportion of patients with minor injuries (ISS = 1-11) who were managed at tertiary centers remained fairly constant at approximately one-third during the first 4 years of the study. This proportion increased sharply and significantly to 42% in 1996-1997 and even more to 77% in 1997-1998 (p < 0.001). The increase in the proportion of patients who were treated at tertiary centers was accompanied by a marked decline in the proportion of patients who were treated at secondary centers. A similar but less dramatic change was observed for the proportion of patients who were treated at primary centers.
Table 13: Distribution of injury severity by center designation and year of study
Table 14: Distribution of injury severity by center designation and phase of regionalization
Figure 3: Proportion of patients treated at tertiary centers.
The proportion of patients with moderate injuries (ISS = 12-24) who were treated at tertiary centers increased significantly, from 56% in 1992-1993 to 82% in 1997-1998 (p < 0.001). Conversely, the proportions of these patients who were treated at primary and secondary centers decreased in an inverse manner.
For patients with major injuries (ISS = 25-49), the proportion treated at tertiary centers increased by a factor of two immediately after the initiation of regionalization, from 36 to 70% in 1992-1993 and 1993-1994, respectively. During the subsequent years, there was a steady and significant increase of this proportion to 84% (p < 0.001). The data show that directly inverse changes occurred in the proportions of patients who were treated at secondary and primary centers.
The majority of patients with fatal injuries (ISS = 50-75) were treated at tertiary centers during all of the years of the study. No significant differences with respect to this proportion were observed during the years of the study or between the phases of regionalization.
The data in Table 15 and Table 16 and Figure 4 describe the prehospital time and time to admission by trauma center designation during the study period. It is important to note that the mean prehospital time decreased linearly and significantly for tertiary, secondary, and primary trauma centers. The interval between discharge from the emergency department and admission to the ward, the intensive care unit, or surgery however, was significantly reduced during the study period in the tertiary trauma centers only (p < 0.001). For primary and secondary centers, there were some fluctuations in this parameter but no significant trends were observed.
Table 15: Prehospital time and time to admission by center designation and year of study
Table 16: Prehospital time and time to admission by center designation and phase of regionalization
Figure 4: Admission delays by trauma center designation and study year.
The data in Table 17 and Table 18 describe the mortality rates by trauma center designation during the study period. These data show that the mortality rate was consistently significantly lower in the tertiary centers than in the secondary and primary centers (p < 0.001). The exception was the 1997-1998 period, when the mortality rate was slightly higher in the tertiary centers compared with the secondary centers. These data also show that the mortality rates in the tertiary and secondary centers declined steadily during the study period. For the primary centers, however, the mortality rate remained fairly constant until the 1997-1998, period when it declined by a factor of two. The increase in the tertiary center mortality and the decrease in the mortality rate in the primary centers during 1997-1998 may be explained by the increase in the number and proportion of patients who were transferred to the tertiary centers. The increase in the proportion of patients with major trauma treated at the tertiary centers may also explain this observation. These explanations are supported by the data in Table 13, Table 14, Table 19, and Table 20. Table 19 and Table 20 show that the proportion of patients transferred to tertiary centers from other hospitals, including primary and secondary centers, increased from 33% in 1992-1993 to 54.3% in 1997-1998. This is an important and statistically significant change (p < 0.001). In addition, the data from the current and previous studies have shown that the morality rate of patients transferred to tertiary trauma centers was higher compared with that among patients who were transported directly to tertiary centers from the scene (Table 21).
Table 17: Mortality by center designation and study year
Table 18: Mortality by center designation and regionalization phase
Table 19: Rates of transfer of patients by year of study
Table 20: Rates of transfer of patients by regionalization phase
Table 21: Multiple logistic regression: impact of trauma system components on mortality risk
The data in Table 22 and Table 23 show that the mean prehospital time was significantly lower for survivors than for nonsurvivors throughout the study period (p < 0.001). The differences with respect to time to admission between survivors and nonsurvivors were not as consistent. The data from the logistic regression analysis showed that after adjusting for the study fiscal year, patient age, and Injury Severity Score, longer prehospital time was associated with a significant increase in the odds of dying (p < 0.001) (Table 21). These data show that for every minute of increase in the prehospital time, the odds of dying increases by 5%.
Table 22: Prehospital and admission delays by status at discharge and by year of study
Table 23: Prehospital and admission delays by status at discharge and by regionalization phase
(Table 24) summarizes the results of a multiple logistic regression used to test the association between the risk of mortality and the phase of regionalization while adjusting for Injury Severity Score and patient age. A number of other models that included mechanism of injury and body region injured as covariates were tested. These more complex models did not produce a significant improvement in the validity, as assessed by the goodness-of-fit statistic. The simpler model, therefore, was retained. The results of this analysis showed that after adjusting for patient age and Injury Severity Score, the process of regionalization produced significant reductions in the risk of mortality. More specifically, compared with the preimplementation period, the adjusted relative odds of dying were 0.39, 0.18, and 0.15 during the initiation, intermediate, and advanced phases, respectively. These estimates are both clinically and statistically (p < 0.001) significant.
Table 24: Multiple logistic regression: impact of trauma care regionalization on mortality risk
The data in Table 25 summarize the results of the multiple logistic regression models evaluating the adjusted association between prehospital time, the trauma center classification, the time to admission, and the method of arrival of the patient to the trauma center (transfer or direct transport). The data show that the relative odds of dying associated with being treated at a tertiary center compared with a secondary center was 0.46. The relative odds of dying associated with being treated at a tertiary center compared with a primary center was 0.25. These data also show that the odds ratio of dying associated with being transferred compared with being transported directly to a secondary or tertiary trauma center was 1.29. Furthermore, although the odds ratio for prehospital time was clinically and statistically significant (p < 0.001), the odds ratio for time to admission was not clinically important or statistically significant (p > 0.05).
Table 25: Multiple logistic regression: impact of trauma system components on mortality risk
A third model was tested in which the variable representing the phase of regionalization was added to those already in the model: prehospital time, trauma center designation, time to admission, and method of arrival at the trauma center. The purpose of testing this model was to estimate the proportion of the variance explained by the regionalization phase that was captured by the variables describing the components of a trauma care system that were included in the second model. The result of this analysis showed that more than 53% of the variance explained by the variable describing the phase of regionalization was attributable to reduction in prehospital time, the trauma center designation, and the method of arrival of the patient at the trauma center (Table 21).
DISCUSSION
This was a prospective cohort study that evaluated the impact of the changes in the trauma care system of Quebec, a Canadian province, on the mortality among patients with major trauma. The study followed the Quebec trauma care system from before to 5 years after the implementation of regionalization. The phases of regionalization were identified by specific landmarks that clearly defined the process of integration of the trauma care services in the province in one system and network. The target population of the study was that of patients with major trauma. The sample was selected after a comprehensive review of all records of patients treated for injuries at the acute-care hospitals of Montreal and Quebec City. The restriction of the sample selection process to these two cities is appropriate for two reasons: first, because these are the two largest urban areas of the province, and second, because the changes in trauma care regionalization were implemented in these locations.
The results of the study demonstrated that since the implementation of the regionalized system, the mortality in major trauma patients was significantly reduced. This reduction in mortality was observed during the initial, the intermediate, and the advanced phases of the regionalization process. It is important to note that the mortality rate continued to decline during each phase and for the entire 5-year period after the introduction of regionalization that was covered by the current study. This decline in mortality remained statistically and clinically significant after adjusting for all potential confounding variables, including patient age and injury severity measures.
The first important component of a trauma care system is the establishment, designation, and support of tertiary trauma centers as the hospitals where patients with severe trauma should be treated. The second important component of a trauma care system is efficient prehospital care, characterized by a short interval between the time of injury and definitive in-hospital care, and by accurate identification and triage of patients with major trauma to tertiary trauma centers. The importance of these components has been generally accepted from a conceptual and a theoretical perspective and has been demonstrated in a limited number of studies. [1,2,30-32] To date, however, the direct relationship between these two components of trauma care systems and reduced mortality has not been empirically demonstrated.
The data from this study showed that during the evolution of the Quebec trauma care system toward regionalization, the prehospital time and the time to definitive care, defined as the time to admission, intensive care treatment, or surgery, were reduced at levels that were both clinically important and statistically significant. In addition, the proportion of patients with major trauma who were treated at tertiary trauma centers was increased. The data also showed that both the prehospital time and the trauma center designation of the treating hospital were important predictors of mortality in this sample of patients with major trauma. More specifically, the risk of mortality increased significantly with longer prehospital times, whereas the mortality risk was lower for tertiary trauma centers compared with secondary and primary trauma centers. These results, therefore, have shown that during the implementation of the regionalized Quebec trauma care system, the two important components of trauma care described above changed significantly and in the appropriate direction and that these changes were associated with significant reductions in trauma-related mortality. Taken collectively, these results have provided empirical evidence that the process of trauma care regionalization, defined as the establishment of trauma centers and of prehospital care aimed at prompt transport of severely injured patients to the appropriate hospital, produces significant benefits by reducing trauma-related mortality. This process-outcome association has been demonstrated in a single system that has been followed during the first 5 years of regionalized trauma care while controlling for the effect of all potential confounding factors.
The results of the stratified analysis showed that implementation of regionalization produced significant reductions in the mortality rates among patients with moderate and major injuries, although the impact on patients with fatal injuries was modest. This is consistent with the principles established by Trunkey stating that the most important impact of trauma care will be realized in the group of patients who have major life-threatening injuries. [1] In these patients, mortality will be reduced if appropriate treatment is provided with minimal delay at tertiary (Level I) trauma centers. In addition, the probability of dying among patients with fatal injuries remains high and in general is not affected by the level of trauma care available.
One of the important observations of the current study was that the time interval from arrival of the patient to the hospital to initiation of definitive care was significantly reduced in the tertiary trauma centers during the process of regionalization. This parameter did not change significantly for the secondary or primary centers. These results may suggest that the reduction in mortality in the tertiary centers may be partially explained by the reduced delay to definitive care. It is equally important to note that the time to definitive care was not lower for the tertiary centers compared with the primary and secondary centers. This observation may suggest that the quality of care available at the tertiary centers may compensate for the longer delays. Nevertheless, this observation also suggests that the emergency rooms of the tertiary centers may be becoming saturated because of the volume of patients and the severity of injuries. In view of this finding, interventions that are aimed at minimizing this saturation should be considered to ensure that these facilities continue to function at their full potential.
The results of the multiple logistic regression models showed that both the prehospital time and the designation level of the treating hospital were independently and significantly associated with the risk of dying. These analyses showed that for every additional minute of prehospital time, the risk of dying increased by 5%. Extrapolation allows us to estimate that the adjusted risk of dying increases by 57% (odds ratio, 1.57), for every 10-minute increase in prehospital time. During the time period covered by the study, the total prehospital time was reduced by an average of 18 minutes. This represents a decrease in the adjusted risk of dying by a factor of 2.25 (odds ratio, 0.44). With respect to the trauma care designation of the treating hospital, the logistic regression models showed that after adjusting for all potential confounders, the risk of dying in tertiary trauma centers was 54% lower than in secondary centers and 75% lower than in primary centers. This is an extremely important observation that further confirms the axiom that patients with major trauma should be treated at tertiary trauma centers. The logistic regression analyses also showed that patients who were transferred to a trauma center had a 30% increase in mortality risk compared with patients who were transported directly from the scene to a trauma center. In the logistic regression models, time to definitive care was not significantly associated with the risk of mortality. This observation suggests that the quality of care of the treating hospital and reduced prehospital times are more important than delays after arrival at the hospital. It is also quite possible that the deleterious effect of long delays to definitive care may be confounded by the designation of trauma centers, because this parameter was significantly reduced in tertiary trauma centers. It is also possible that in tertiary trauma centers patients with severe injuries are properly and promptly triaged after arrival at the hospital, thus reducing the delay to definitive care for this high-risk group. Conversely, in the tertiary centers, patients with less severe injuries that do not require immediate attention may have longer waiting times to definitive care. The increased volume of patients with such injuries may explain the higher delay to definitive care but reduced mortality in tertiary centers.
The final logistic regression model showed that the changes in prehospital time, trauma center designation, method of arrival of the patient to the trauma center, and time to definitive care explain more than 50% of the variance in mortality captured by the process of regionalization of trauma care. This is an important observation because it provides direct evidence that these components of a trauma care system are the major factors contributing to its effectiveness in reducing mortality. This observation also suggests that a proportion of the effect of trauma care regionalization may not be explained by changes in these components of the system. Other factors may contribute to the effectiveness of trauma care regionalization in Quebec that are not captured by these variables because they are more universal in nature. These factors may be changes in prehospital care, increased awareness and education regarding the management of trauma patients at all acute-care hospitals, and improved technology that has become available to all trauma hospitals during the period covered by the study. In addition, it should be pointed out that from a statistical viewpoint, explanation of more than 50% of the variance in the outcome variable is rare, indicating an extremely large effect size.
The strengths of the present study include its prospective design and the large sample size that ensured sufficient power to conduct important subgroup comparisons and analysis with reliable and valid results. The study followed without interruption the evolution of the Quebec trauma care system from before implementation to the advanced stages of regionalization. The data were collected from all acute-care hospitals that treated trauma patients from the two urban areas where regionalization was introduced. The comprehensive coverage of patient types and hospitals eliminates any selection bias. Data were available on all potential confounders, and statistical analyses were used to control their effects on the study results.
The current study produced direct empirical evidence to support the benefits of an integrated regionalized trauma care system. The independent benefits of tertiary trauma centers and reduced prehospital time were also clearly demonstrated. The results are compatible with those that have been reported previously in the literature. The current study, however, has used data from a single region to demonstrate the process-outcome relationship of trauma care regionalization. Other authors have supported this relationship based on theoretical concepts and the accumulation of evidence from a number of studies, each one evaluating a specific component of the trauma care system.
The conclusions of the present study are the following. First, trauma care regionalization produces significant reductions in the mortality of patients with major trauma. Second, the designation of tertiary trauma centers where patients with severe injuries should be treated and reduced prehospital times are both essential if these benefits are to be realized. Further research should be conducted that will compare different models of trauma care systems and that will identify other components of trauma care that could be improved. The impact of trauma care systems on patient outcomes other than mortality, including disability, morbidity, and quality of life, should be evaluated. Finally, there is an immediate need for a complete economic evaluation that will include cost-effectiveness assessments of trauma care systems.
DISCUSSION
Dr. A. Brent Eastman (La Jolla, California): Dr. Sampalis and his colleagues in Montreal are to be doubly congratulated, for not only presenting a very important paper, but for the establishment of a trauma system in the province of Quebec. Dr. Mulder particularly was instrumental in the development and evolution of this trauma system.
The authors had the foresight to plan and initiate a prospective cohort study coincident with the creation of this trauma system. This was done for the express purpose of proving the relationship between certain components of their trauma system and a decrease in mortality rate. They have succeeded by documenting a threefold decrease, 52 to 18%, in the mortality rate of trauma patients.
As with many comprehensive studies, this study raises more questions than it answers. You present this as an analysis of the Quebec Regionalized Trauma System, and yet the study sample is clearly "patients admitted to the acute-care hospitals in Montreal and Quebec City," not the entire province. Since one of the principal components examined was prehospital times, it is critical to know what effect travel distances and weather conditions had on this component.
You do not specifically address such factors as prehospital airway control. Was there any change in prehospital intubation protocols during this study period? The mortality related to prolonged prehospital times may be related to airway problems, and if the treatment of this was improved during the study period, that may also have contributed to improved survival.
You make reference to overall increased survival when major trauma patients are taken expeditiously to designated tertiary hospitals. However, you did not specifically define if there was a decrease in mortality rates in these tertiary hospitals per se after implementation of the system. For example, did the survival rate increase at McGill University after implementation of the system?
One of the most pressing questions in the U.S. today is the relationship between volume, either surgeon or center, and outcome. This is being specifically addressed in the current version of the ACSCOT document for optimal care of the injured patient. Your database may provide some answers on this critical question. Did your tertiary centers realize an increased volume with regionalization, and was this a factor in improved survival rates?
One of the perennial problems in studying the issue of prehospital triage has been the absence of data from nontrauma hospitals, therefore the lack of a denominator. I understand from personal communications with your coauthor, Dr. Mulder, that with your single-payer system you've been able to tie hospital reimbursement to the provision of trauma data and that you do have data from all acute-care hospitals. With those data, you will be able to accurately determine the sensitivity and specificity of your prehospital triage instrument.
Given the importance that most of the trauma world places on the efficacy of the ATLS, it was notable that you do not mention it in your manuscript. You found an increase in ISS, motor vehicle crashes, and injuries to the head, thorax, and abdomen. This seems counterintuitive, given the increased use of seat belts, airbags, and drunk-driving prevention efforts. Can you explain?
I'm interested that you did not see a significant change in survival of the most severely injured patients. Many systems have found this subset to be where the greatest improvement was found.
This important paper will be frequently quoted and will provide powerful information for those attempting to overcome the political and economic barriers to creating much-needed trauma systems. I would close by asking two questions. Do your data provide evidence that outcome is improved with increased volume? And two, is prevention an essential element of your system? Only prevention will address the immediate deaths that you, in your manuscript, call nonpreventable. I would submit that these immediate deaths represent the greatest opportunity for prevention. I thank the Association and the authors for the opportunity of discussing this important paper.
Dr. Ari K. Leppaniemi (Helsinki, Finland): Finland doesn't have any certified trauma centers at the moment, and we need to review our trauma care system. My question is, since the size of Finland is about the same as Quebec is with a little smaller population, are four tertiary trauma centers for 8 million people in that land area sufficient, or should there be maybe less?
Dr. William B. Long (Portland, Oregon): The Oregon trauma system has been operational for 10 years, but we still have problems with the Oregon Trauma Register getting completeness of data. In fact, until this year, we had 35% missing data from the rural hospital Level IIIs and IVs. I would be interested to know what percentage of your Levels IIs, IIIs and IVs in Quebec were contributing data and how complete it was. Thank you.
Dr. Richard J. Mullins (Portland, Oregon): Are patients with head injury the group that seems to have benefited the most from implementation of the trauma system in Quebec? If that is the case, Dr. Sampalis, do you have an opinion what are the specific reasons why head-injured patients have an improved outcome when they are treated in a trauma system?
Dr. Lawrence H. Pitts (San Francisco, California): Dr. Sampalis, you mentioned that the number of secondary centers had fallen slightly as the various periods went on, and that this was better matched to population. How was that reduction achieved? Was it voluntary cessation of care by some of the centers, or was that induced by some other method?
Dr. John S. Sampalis (closing): Thank you very much for the questions. I'm going to try to be brief. With respect to the comments raised by Dr. Eastman regarding prehospital travel time and prehospital care, I have to mention that in Quebec the situation with respect to prehospital trauma care is quite different than what exists in the United States.
We have two types of trauma care: one that's available in Montreal, where physicians may be dispatched to the scene to provide Advanced Life Support; and the other kind of prehospital care that we have is essentially what we hope is right now scoop-and-run, which is Basic Life Support provided by emergency medical technicians. Therefore, airway management at the scene is restricted to a small number of patients that are treated by physicians in Montreal.
I'm not sure whether we are observing a circular trend in hospitals doing better overall. I think that's an interesting question. However, if I am to try and assess what is happening with respect to our trauma patients, I think that in fact we are treating these patients better.
This is a study that has followed the series of other studies where we first showed that, in fact, in the tertiary trauma centers after designation, the mortality was reduced. Therefore, it follows from these other studies that we are actually treating trauma patients better in the entire system.
Did tertiary trauma centers realize an increase in volume? I think this is accurate to say, as we can see by the increase in the proportion of patients with major trauma that were treated at these institutions.
I did not mention ATLS, and that's a very important point. However, from what I understand from our system, most of the surgeons practicing in tertiary trauma centers have taken this training.
The point that there has been an increase in motor vehicle collisions might be counterintuitive given the existence of preventive measures and airbags and seat belts, etc. However, the fact that motor vehicle accidents do occur has nothing to do with the availability of a seat belt or airbag. Therefore, these patients continue to come to us. In fact, we have seen an increase, a change in the profile of trauma in our population, with a younger population, more and more vehicle accidents, and more violent crimes.
Do the data provide evidence that the outcome has improved in the entire system? I think so. And the fact that we use data from only Montreal and Quebec City is explained by the point that these are the two major cities in the province. This is where most trauma patients are being treated, and the implementation of the trauma care system at this point in time focused on these two cities. Therefore, I wanted to see what was happening in these two regions before starting to compare the rest of the province.
I think the next phase of studies that we will do with the expansion of the network to the entire province will evaluate the impact on the other hospitals. Are four tertiary trauma centers necessary or enough for 8 million inhabitants? I'm not sure whether I can answer this right now. This is probably an administrative question that we will have to address very soon in the province.
Completeness of data. That's a very important point. I should mention that we didn't only rely on the Trauma Registry, but this study was also funded by the government of Canada. Therefore, any data that we were missing in the registry, we went and captured from the hospital charts. We included patients that we had most of the information available that we needed.
Are patients with head trauma doing better? I assume they are, and that's a very important point that we need to conduct a subanalysis.
Why were the number of secondary centers reduced? This was done by a Committee of Trauma Care in the province that evaluated the performance of these centers and the need to have them within the network. It was decided that they weren't necessary, given their situation and the volume of patients that they treated. This volume of patients now would go to the other secondary and the tertiary centers. Thank you very much for the discussion and the privilege of presenting the paper.
