Journal of Trauma and Acute Care Surgery:
AAST Plenary Papers
The epidemiology of trauma-related mortality in the United States from 2002 to 2010
Sise, Robert G. MBA, MPH; Calvo, Richard Y. PhD(c); Spain, David A. MD; Weiser, Thomas G. MD, MPH; Staudenmayer, Kristan L. MD, MS
From the Medical School (R.G.S.), University of California, San Francisco; and Trauma Service (R.Y.C.), Scripps Mercy Hospital, San Diego; and Department of Surgery (D.A.S., T.G.W., K.L.S.), Stanford University, Stanford, California.
Submitted: September 22, 2013, Revised: December 10, 2013, Accepted: December 17, 2013.
This study was presented at the 72nd annual meeting of the American Association for the Surgery of Trauma, September 18–21, 2013, in San Francisco, California.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML text of this article on the journal’s Web site ( www.jtrauma.com).
Address for reprints: Kristan Staudenmayer, MD, MS, Department of Surgery, Stanford University, 300 Pasteur Dr, Grant Bldg, S-067, Stanford, CA 94305; email: email@example.com.
Epidemiologic trends in trauma-related mortality in the United States require updating and characterization. We hypothesized that during the past decade, there have been changing trends in mortality that are associated with multiple public health and health care–related factors.
Multiple sources were queried for the period of 2002 to 2010: the National Trauma Data Bank, the National Centers for Disease Control, the National Highway Traffic Safety Administration, the Nationwide Emergency Department Sample, and the US Census Bureau. The incidence of injury and mortality for motor vehicle traffic (MVT) collisions, firearms, and falls were determined using National Centers for Disease Control data. National Highway Traffic Safety Administration data were used to determine motor vehicle collision information. Injury severity data were derived from the Nationwide Emergency Department Sample and National Trauma Data Bank. Analysis of mortality trends by year was performed using the Cochran-Armitage test for trend. Time-trend multivariable Poisson regression was used to determine risk-adjusted mortality over time.
From 2002 to 2010, the total trauma-related mortality decreased by 6% (p < 0.01). However, mortality trends differed by mechanism. There was a 27% decrease in the MVT death rate associated with a 20% decrease in motor vehicle collisions, 19% decrease in the number of occupant injuries per collision, lower injury severity, and improved outcomes at trauma centers. While firearm-related mortality remained relatively unchanged, mortality caused by firearm suicides increased, whereas homicide-associated mortality decreased (p < 0.001 for both). In contrast, fall-related mortality increased by 46% (5.95–8.70, p < 0.01).
MVT mortality rates have decreased during the last decade, owing in part to decreases in the number and severity of injuries. Conversely, fall-related mortality is increasing and is projected to exceed both MVT and firearm mortality rates should current trends continue. Trauma systems and injury prevention programs will need to take into account these changing trends to best accommodate the needs of the injured population.
LEVEL OF EVIDENCE
Epidemiologic study, level III.
It is important to follow epidemiologic trends in trauma-related mortality to allocate resources appropriately. Recent reports have either focused on a single mechanism (e.g., firearm-related injuries),1 specific injury patterns (e.g., traumatic brain injury),2 comorbidities (e.g., obesity),3 or disparities (e.g., payer status)4–6 or confined their analysis of the epidemiology of trauma deaths in a single region or trauma center.7,8 In contrast, the relevant public health and policy literature focuses on broader trends in injury or specific public health issues such as firearms.9,10 There has been no recent study that evaluates national mortality trends and their associated causes for those mechanisms that are frequently managed at trauma centers.
The current study aimed to explore the mortality trends and associated causes for the most common mechanisms of injuries treated at US trauma centers during the past decade. Since the contribution to final mortality estimates span multiple phases including injury prevention, the prehospital setting, and health care facilities, we used multiple sources of data to create a complete picture of mortality. The aims of this study were to describe the overall trend in trauma-related annual mortality in the United States from 2002 to 2010 and to determine how trauma mortality rates are impacted by mechanism-specific characteristics.
PATIENTS AND METHODS
Multiple sources of data were used because no single source contains complete information on all phases of care (e.g., scene vs. hospital). A brief description of the data sources is presented later. A detailed description of the data sources and methods can be found in Appendix A (Supplemental Digital Content 1, http://links.lww.com/TA/A368).
National rates for mechanism-specific injuries were obtained using the Web-based Injury Statistics Query and Reporting System (WISQARS),11 which is a publically available data source made available by the National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. WISQARS provides data for both nonfatal and fatal injury rates. Detailed information on motor vehicle traffic (MVT) collisions was obtained through the publically available reports issued by the National Traffic Highway Safety Administration (NHTSA). There are two data systems available from NHTSA: the National Automotive Sampling System General Estimates System and the Fatality Analysis Reporting System. The General Estimates System was used to determine the annual vehicle miles traveled, total number of collisions, and total number of persons who sustained injuries from MVT collisions. The Fatality Analysis Reporting System was used to derive 30-day mortality rates based on the type of MVT collision.
National rates of emergency department visits and admission rates were determined using data derived from the Agency for Healthcare Research and Quality’s Healthcare Cost and Utilization Project (HCUP), Nationwide Emergency Department Sample (NEDS). NEDS is a nationally representative sample of all emergency department visits in the United States. Data collection began in 2006 and is available through 2010. NEDS data contain trauma-specific information such as trauma center status, mechanism of injury, and injury severity and were queried using the HCUP’s Web-based tool, HCUPnet.12 The National trauma Data Bank (NTDB) from the American College of Surgeons from 2002 to 2010 was used to determine outcomes at US trauma centers. US Census Bureau data were reviewed to discern the change from 2000 to 2010 in the percentage of the US population that is 60 years and older.13
Unadjusted analyses of mortality trends were performed using the Cochran-Armitage (CA) test for trend. Mortality rates were presented as rate per 100,000 US population derived using US Census population data. Annual mean Injury Severity Scores (ISS) were derived from the NTDB and were analyzed between 2002 and 2010 with t tests.
Time-trend multivariable Poisson regression was used to evaluate adjusted mortality over time for hospital outcomes (using NTDB and NEDS). Variables included in the modeling procedure were categorized age, race, insurance status, mechanism of injury, sex, ISS, and the number of comorbidities. Race was classified as white (Hispanic and non-Hispanic), Black (including African American), Asian (including Pacific Islander), American Indian (including Alaskan Native), other, or unknown. Preexisting comorbidities analyzed in the NEDS data set were based on the Charlson Comorbidity index.14 Comorbidities used in the NTDB analyses were based on 46 comorbidity classes collected by the American College of Surgeons’ Committee on Trauma NTDB system and are mappable to the Charlson comorbidity index.15 The number of trauma centers in the NTDB increased every year, so the total patient load per year was used as the offset variable. Clustered robust SE estimation was used with medical facility as the grouping variable. Akaike’s information criterion values were used to select the best model among a series of candidate models containing different variables. Akaike’s information criterion values are a transformation of the maximum likelihood function used to generate model estimates. Over dispersion of Poisson-distributed data was evaluated using the goodness-of-fit test under a χ2 distribution and with a p < 0.200. For all final models, there was no evidence of lack-of-fit or over dispersion. Results are presented as incident risk ratios (IRRs).
Data were analyzed using Stata/MP version 11.2 (StataCorp LP, College Station, TX) and R Statistical Software (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was attributed to p < 0.05.
Comparison of Trends: NEDS Versus NTDB
There were two sources for in-hospital mortality and patient-level variables in the current study (NTDB and NEDS). Because these two sources differ in data collection methodologies, it is necessary to compare patient characteristics between the two to validate observed trends. NEDS data are only available from 2007 to 2010, so comparisons between the NEDS and NTDB were only conducted for these years. To determine if both data sets experienced similar mortality trends, we generated estimates of incidence relative risk for mortality using Poisson regression for the years of 2007 to 2010 for both NTDB and NEDS data sets using common variables. Risk estimates were subsequently compared in magnitude, direction, and statistical significance to evaluate if patients in both data sets experienced similar levels of risk for death (see findings in Appendix B, Supplemental Digital Content 2, http://links.lww.com/TA/A369).
WISQARS data were used to determine injury-related deaths during the study period. Of the trauma-related deaths, 23% (365,090) were caused by unintentional MVT mechanisms, 12% (190,040) were caused by unintentional falls, and 17% (266,065) were caused by firearms (homicide and suicide). Homicide by cutting or piercing constituted only 1% and burns only 2% of all injury-related deaths (18,021 and 28,278 deaths, respectively). Other injury-related deaths were caused primarily by medical causes such as poisoning, suffocation, drowning, and adverse effects. Since MVTs, falls, and firearms were the most frequent mechanisms of injury-related mortality associated with trauma center care, we focused the analysis on these three mechanisms.
Injury-related mortality caused by MVT, falls, and firearms decreased in the United States from 2002 to 2010 (from 32.2 to 30.5 deaths per 100,000 US population, p < 0.001 per CA test for trend). However, the trends for each of the mechanisms differed (Fig. 1). The mortality rate for MVTs decreased from 2002 to 2010 (15.6 to 11.5 deaths per 100,000 US population, p < 0.001 per CA test for trend). In contrast, mortality rates increased for fall-related mechanisms during the same period (5.9–8.7 deaths per 100,000 US population, p < 0.001 per CA test for trend). Firearm-related mortality rates (including both suicide and homicide) decreased slightly from 10.5 to 10.3 (p =0.03 per CA test for trend).
The category of MVT includes all traffic-related mechanisms (i.e., pedestrian, motorcycle collisions, and injuries to vehicle occupants). We used data from NHTSA to determine the frequency and trends associated with MVT-related mechanisms. The mechanism that accounts for the largest portion of MVT-related fatalities is death of vehicle occupant (75%). The remainder of MVT-related mortality includes pedestrian (12%), motorcycle (11%), and other forms of land transport (2%) deaths. The trends of these MVT mechanisms differed slightly. Motorcycle-related mortality increased slightly from 2002 to 2010 (from 1.1 to 1.5 per 100,000 US population, p < 0.001 per CA test for trend), while both vehicle occupant–related mortality rates and pedestrian deaths decreased. Vehicle occupant deaths decreased by 36% (11.8 to 7.5 deaths per 100,000 US population, p < 0.001 per CA test for trend), and pedestrian-related mortality rates decreased by 18% (1.7 to 1.4 deaths per 100,000 US population, p < 0.001 per CA test for trend). Since occupant-related mortality constituted the greatest portion of MVT deaths, we focused further analysis on this subgroup. To determine if improvements in vehicle occupant mortality were caused by changes in the number of motor vehicle collisions, fewer injuries, less severe injuries, or improved medical care, we compared the trends of each of these components using various data sources.
NHTSA data were also used to determine the rate of motor vehicle collisions. From 2002 to 2010, the rate of motor vehicle collisions in the United States decreased 20% (from 2,196 to 1,752 per 100,000 US population, p < 0.001 per CA test for trend, Fig. 2), and the number of car occupant injuries per collision decreased by 19% (from 28.6 to 23.1 car occupants injured per 100 motor vehicle collisions, p < 0.001 per CA test for trend). The decrease in the number of collisions was not associated with a large decrease in vehicle miles traveled during that time (3% decrease, from 9,930 to 9,592 vehicle miles traveled per US population).
HCUP’s NEDS data on emergency department visits were used to determine if the rate of visits to US emergency departments changed. Data were available from 2006 to 2010. During this period, there was a 7% decrease in the number of emergency department visits for occupants of motor vehicle collisions (from 959 to 887 patients per 100,000 US population, p < 0.001 per CA test for trend). This was also associated with a lower admission rate for occupants of motor vehicle collisions. In 2006, 6.6% of motor vehicle occupants seen in US emergency departments were admitted (189,523 of 2,860,464 visits), compared with 5.7% in 2010 (156,495 admissions of 2,737,646 visits, p < 0.001 per CA test for trend).
To determine if injury severity had similarly decreased, we used NEDS to determine the ISSs of occupants of motor vehicle collisions seen in US emergency departments. Complete information for the NEDS patient population was available from 2007 through 2010. During this time, the mean ISS for occupants of motor vehicle collisions seen in US emergency departments decreased slightly (2.5 to 2.4, p < 0.001 per t test), but the percentage of severely injured patients as measured by an ISS greater than 15 did not significantly change (1.8% to 1.7%, p =0.99 per t test). Injury severity was compared for occupants of MVT collisions who were admitted to trauma centers and nontrauma centers from 2007 to 2010 (Fig. 3). Overall, injury severity was higher at trauma centers than at nontrauma centers, but the trends between the two were similar. There was a decrease in the mean ISS in both types of hospitals from 2007 to 2009, followed by an increase in injury severity in 2010.
To determine whether risk-adjusted outcomes at trauma centers improved for occupants of motor vehicle collisions during the study period, multivariable modeling was performed. After adjusting for age, sex, injury severity, comorbidities, and insurance status, risk-adjusted mortality decreased each year at trauma centers (Table 1). The IRR for vehicle occupants in 2010 was 23% lower relative to 2002 (IRR, 0.77 in 2010 compared with 2002).
Fall-related mortality increased 46% from 2002 to 2010. To determine the association between age and risk for fall-related death, we analyzed the mortality of fall-related death per 100,000 US population in 2010 using WISQARS data (Fig. 4). Fall-related death was associated with advanced age with rates rapidly increasing during the age of 60 years. While fall-related mortality is increasing, so is the burden of injury caused by nonfatal falls. There was a 23% increase in the number of fall-related injuries from 2002 to 2010 (from 7,438,416 to 9,164,291, p < 0.001 per CA test for trend). Overall, the combined rate of both fatal and nonfatal fall-related injuries has increased from 2,592 to 2,977 per 100,000 age-specific US population.
To determine if the increase in mortality was associated with differences in hospital mortality at trauma centers, we used the NTDB to determine adjusted mortality controlling for age, sex, injury severity, comorbidities, and insurance status (Table 1). During the study interval, the average adjusted risk of mortality decreased with each year (IRR, 0.96; 95% confidence interval, 0.95–0.97; p < 0.001). At the end of the study period (2010), the annual decrease in adjusted mortality resulted in a 22% reduction in the risk of death compared with 2002 (IRR, 0.78).
Using data from NEDS, we determined that approximately 63% of all firearm-related injuries were seen in trauma centers during 2006 to 2010. For assaults and homicides, 78% were cared for at trauma centers compared with 53% of self-inflicted firearm injuries. We then compared trends versus suicide-related and homicide-related mortality using WISQARS (Fig. 5). Firearm-related mortality caused by suicide increased, whereas homicide-associated mortality decreased. These trends were both statistically significant (p < 0.001 per CA test for trend for both).
To determine if the lack of improvement in mortality seen for all firearm injuries was mirrored by a lack of mortality benefit over time at trauma centers, we used the NTDB to determine adjusted mortality. While national rates of firearm-related mortality were relatively unchanged from 2002 to 2010, results from multivariable analyses revealed an increased risk for mortality during the study duration in trauma centers (Table 1). After adjustment, patients who were involved in firearm-related injury were more likely to experience in-hospital mortality in each successive year (IRR, 1.02; 95% confidence interval, 1.00–1.04; p =0.014).
Annual injury-related mortality rates decreased from 2002 to 2010, but there were striking differences in mortality trends between mechanisms. The observed decrease in mortality was caused by a 27% decrease in MVT deaths, which was partially offset by a 46% rise in fall-related mortality. If current trends continue, fall-related deaths will exceed MVT deaths as the most common cause of trauma-related mortality by 2013 to 2014. In contrast, firearm-related mortality improvements were small and changed relatively little compared with the other two mechanisms. The differences in the magnitude and direction of these trends argue that trauma-related mortality is affected by multiple factors.
Improvements in MVT deaths were caused by improvements on multiple fronts including crash avoidance, car safety, and medical care. From 2002 to 2010, there was a large decrease in the number of motor vehicle collisions (20%) and in the number of injuries per collision (19%). The decrease in the number of occupant injuries translated into fewer visits to US emergency departments, lower rates of admission, and less severe injuries. Improvements in occupant mortality were also caused by improvements in hospital care. In trauma centers, there was a 23% decrease in the adjusted risk for death from 2002 to 2010. While we were unable to determine if nontrauma centers experienced a similar trend, comparisons between the NTDB and NEDS from 2007 to 2010 suggest the trend seen in the NTDB may reflect trends occurring at nontrauma centers as well (Appendix B, Supplemental Digital Content 2, http://links.lww.com/TA/A369).
In contrast, fall-related mortality substantially increased. Falls are exponentially associated with increasing age and during the study period, the number of US citizens older than 70 years increased by 2.4 million. However, fall-related mortality is not simply a function of age. When the number of nonfatal falls and fatal falls were normalized to age-specific US population numbers, the rate of fall-related injuries still increased (from 2,592 to 2,977 per 100,000 age-specific US population). Fall syndromes are increasingly studied and are known to be multifactorial including patient and environmental factors.16,17 It is possible that as the population ages, greater number of elderly patients are forced to live in environments that are unsafe relative to their functional status. It may also be that the outpatient health care system is unable to absorb the large number of complex elderly patients, and as a result, patients are not receiving optimal medical care. Despite these disturbing trends, it seems that there is an improvement in the hospital mortality rates for fall-related injuries at trauma centers. There is approximately a 4% reduction each year in adjusted mortality, even after controlling for all known confounders.
Trends in firearm-related deaths were less dramatic. While the overall trend in firearm-related mortality is relatively unchanged, there were differences in the trends for suicides versus homicides. Suicide-related deaths consistently exceeded homicide deaths for the duration of this period. Furthermore, the rates for both trends were relatively unchanged until 2006, after which suicide-related mortality increased while homicide-related mortality decreased. The trends in suicide and homicide obtained in the current study are consistent with other reports covering the period from 2006 to 2010 and have been attributed to changes in economics, demographics, policing, and policy.10 While trauma center care seemed to improve over time for fall and motor vehicle occupant injured patients, the opposite was true in the case of firearm-related injuries. The reason for this is not clear. Since the other two mechanisms fared better in later years at trauma centers, it would be unlikely that differences in outcomes after firearm-related injuries were caused by worsening care and suggest that there is a change in the nature of firearm injuries presenting to trauma centers. For example, it may be that the increase in mortality is partially caused by the changing pattern of firearm injuries in the United States. Since suicide injuries produce higher mortality rates, it is possible that the relative increase in suicide-related injuries is driving this change.
The current study has some limitations. The sampling methodologies of each database varies, which may affect the comparability of results between data sets. For example, WISQARS draws mortality information from the national vital statistics system, whereas the NTDB and NEDS mortality information are derived directly from hospitals. Furthermore, longitudinal analysis is complicated by varying techniques in data collection over time. The NTDB changed data collection standards from 2006 to 2007. The direction of mortality trends from 2002 to 2006 and 2007 to 2010 were consistent, suggesting that while specific numbers may be different, the trends were still informative. In addition, the NTDB and the NEDS only contain patients who were admitted to US trauma centers and emergency departments, respectively. Although both are highly statistically powered, estimates derived from these data sets may overestimate the effects of certain risk factors owing to the possibility of a proportion of the injured populations who do not seek treatment. Furthermore, NEDS data were only available from 2007 to 2010, which limits the ability to derive relevant estimates for the earlier years and determine broader trends. Finally, the effect of comorbidities was difficult to assess owing to differences in reporting of certain conditions over time by participating centers. Future research efforts should seek to evaluate the effect of the growing elderly trauma population and effects of chronic conditions on clinical outcomes.
Patterns of trauma-related mortality have changed substantially during the past decade. The combined efforts of the US government, the automobile industry, safety organizations, and the health care system have resulted in marked improvements in deaths caused by MVT collisions. However, an increasingly elderly population has presented new challenges with a dramatic increase in fall-related injuries, which threatens to exceed the burden of injury from MVT collisions and firearms. In contrast to the public health success represented by decreasing MVT deaths, firearm-related mortalities remain comparatively unchanged. We will need to adjust trauma resources and prevention efforts to accommodate the changes in trauma-related mortality and changing US demographics.
R.G.S., R.Y.C., and K.L.S. performed the literature search, statistical analysis, data interpretation, and writing of the manuscript. K.L.S., D.A.S., and T.G.W. performed the study design, reviewed the statistical analysis, and performed the critical revision of the manuscript.
The authors declare no conflicts of interest.
1. Sauaia A, Miller JI, Moore EE, Partrick D. Firearm injuries of children and adolescents in 2 Colorado trauma centers: 2000–2008. JAMA. 2013; 309: 1683–1685.
2. Alkhoury F, Courtney J. Outcomes after severe head injury: a National Trauma Data Bank–based comparison of level I and level II trauma centers. Am Surg. 2011; 77: 277–280.
3. Brown CV, Neville AL, Rhee P, Salim A, Velmahos GC, Demetriades D. The impact of obesity on the outcomes of 1,153 critically injured blunt trauma patients. J Trauma. 2005; 59: 1048–1051;
4. Haider AH, Chang DC, Efron DT, Haut ER, Crandall M, Cornwell EE 3rd. Race and insurance status as risk factors for trauma mortality. Arch Surg. 2008; 143: 945–949.
5. Greene WR, Oyetunji TA, Bowers U, Haider AH, Mellman TA, Cornwell EE, Siram SM, Chang DC. Insurance status is a potent predictor of outcomes in both blunt and penetrating trauma. Am J Surg. 2010; 199: 554–557.
6. Englum BR, Villegas C, Bolorunduro O, Haut ER, Cornwell EE 3rd, Efron DT, Haider AH. Racial, ethnic, and insurance status disparities in use of posthospitalization care after trauma. J Am Coll Surg. 2011; 213: 699–708.
7. Cothren CC, Moore EE, Hedegaard HB, Meng K. Epidemiology of urban trauma deaths: a comprehensive reasessment 10 yers later. World J Surg. 2007; 31: 1507–1511.
8. Kahl JE, Calvo RY, Sise MJ, Sise CB, Thorndike JF, Shackford SR. The changing nature of death on the trauma service. J Trauma Acute Care Surg. 2013; 75: 195–201.
9. Bergen G, Chen LH, Warner M, Fingerhut LA . Injury in the United States: 2007 Chartbook. Hyattsville, MD: National Center for Health Statistics; 2008; .
10. Centers for Disease Control and Prevention. Violence-related firearm deaths among residents of metropolitan areas and cities—United States, 2006–2007. MMWR Morb Mortal Wkly Rep. 2011; 60:(18): 573–578.
11. National Center for Injury Prevention and Control (NCIPC). Web-based Injury Statistics Query and Reporting System (WISQARS). Available at: http://www.cdc.gov/injury/wisqars/index.html
. Centers for Disease Control and Prevention. Atlanta, GA. Accessed August 1, 2013.
12. HCUPnet. Healthcare Cost and Utilization Project. Available at: http://hcupnet.ahrq.gov
. Agency for Healthcare Research and Quality, Rockville, MD. Accessed August 1, 2013.
14. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987; 40: 373–383.
16. Deandrea S, Lucenteforte E, Bravi F, Foschi R, La Vecchia C, Negri E. Risk factors for falls in community-dwelling older people: a systematic review and meta-analysis. Epidemiology. 2010; 21: 658–668.
17. Gerson LW, Camargo CA Jr, Wilber ST. Home modification to prevent falls by older ED patients. Am J Emerg Med. 2005; 23: 295–298.
Dr. Christopher C. Baker (Roanoke, Virginia): Dr. Sise and his coauthors have analyzed mortality from trauma in the U.S. from 2002 to 2010, focusing mainly on deaths due to motor vehicle crashes, firearms and falls.
They have utilized several large databases from different sources using a complex statistical modeling. These data are instructive, showing a decrease in motor vehicle crashes and MVC mortality over time which they attribute to increased safety of the vehicles and trauma center care.
Conversely, they showed an increase in the fall-related deaths, particularly in patients older than 65. These data clearly have profound public health implications. I have four questions for the authors.
In your incidence risk ratio did you consider using the Mantel-Haenszel estimate statistic? Secondly, why did you use ICD-10 codes since these are not yet fully imbedded in most hospitals?
Third, how do you explain the increased mean comorbidity count in the NEDS database compared to the NTDB database?
Finally, what recommendations would you make for impacting the increase in fall-related mortality rates in the elderly?
I commend the authors on an excellent study and thank the association for the privilege of the floor.
Dr. Frederick Moore (Gainesville, Florida): In 2011, Arbabi and colleagues from Harborview published an interesting study in JAMA. They performed a retrospective cohort study of over 120,000 adult trauma patients entered into the Washington State Trauma Registry from 1995 to 2008. They showed that in-hospital mortality had decreased substantially over this time period. However when you look at mortality at 1 year post discharge, it did not change. This is occurring because we are discharging more patients to long-term acute care (LTAC) facilities where they just die an indolent death. If you look at the Glue Grant data you see the same signal. They looked at time to recovery from organ dysfunction after severe blunt trauma. At 14 days, 37% of the patients still in the ICU and had significant organ dysfunction. It is not clear what happened to these patients, but I assume, based on my recent experience in Gaineville, many were discharged to LTACs and experienced dismal long-term outcomes. In the future we need to define chronic critical illness after severe blunt trauma and have a better understanding of what happens to these patients. As the population ages, this is an epidemic that is not going to go away.
Mr. Robert G. Sise (Stanford, California): Thank you very much for those great points and questions.
In regard to our statistical analysis, we performed a Poisson regression which in many ways is similar to your basic logistic regression, however, the focus is instead on measuring a variety of different outcomes, focusing on generating one type of count. That was the focus in this analysis.
In terms of detailed questions about what was included and what was excluded for various options, the logistic regression analysis versus Poisson, I will defer to some of my coauthors to address that question at a later point in time.
In terms of our selection of ICD-9 versus ICD-10 codes, that was mainly a function of the datasets. We standardized those codes across datasets using some well-known and well-respected conversion tables.
In terms of what we can do to address fall mortality at our trauma centers, this is a true challenge for the trauma system given that a lot of this is driven by the shifting demographic or our aging population. In many cases fall mortality is a proxy for the frailty of patients arriving at trauma centers.
This really speaks to how we will address elderly patients presenting with multiple comorbidities at trauma centers. Should we somehow activate the trauma system differently? Channel them into other services? Are they better suited for medicine? This is all something that we will have to tackle over the years to come.
In terms of the reference about potentially some mortality not being captured by the trauma registry because patients are being shifted over to long-term care facilities or skilled nursing facilities, the reason we used the WISQARS dataset as well as NHTSA dataset was to capture some of those long-term outcomes, expanding well beyond discharge from trauma centers.
I would like to thank the Association for the privilege of having the floor.
Trauma mortality; epidemiologic trends; motor vehicle traffic collisions; firearms; falls
Supplemental Digital Content
Copyright © 2014 by Lippincott Williams & Wilkins
Highlight selected keywords in the article text.