As the US population ages, elderly persons account for an increasing number of trauma patients. Morbidity and mortality from traumatic injury are higher in elderly patients compared with younger patients.1- 2 It is well known that the care found in trauma centers benefits appropriately triaged patients3- 6 and is reserved for the most severely injured patients.7
Early studies suggest that elderly patients may be less likely to be admitted to a trauma center after an injury than are younger patients.8- 9 Decreased referral to trauma centers may be related to a lack of referral centers, but elderly patients would not be affected disproportionately if this were the case. Alternatively, decreased referral may result because standard trauma triage criteria are inappropriate or less accurate in older patients.10- 14
Few population-based data describe the public health effect of trauma in elderly patients or how often elderly patients receive care at a trauma center. Existing studies are either single-site studies or do not account for system-level factors known to influence likelihood of admission to a trauma center, specifically proximity to a trauma center and trauma center availability within county of residence. Trends of longer life spans and an aging baby boomer population forecast a doubling of older American adults to 71 million in the next 25 years (20% of the US population). It is therefore critical to understand current trends in the care of injured elderly patients.15
We analyzed data from the state of California to determine the likelihood of admission to a trauma center for injured elderly persons after adjusting for patient risk factors and hospital and system-level factors. Different trauma types have different prognoses that could account for differences in admission rates, so we also considered likelihood of admission to trauma center in these groups by injury type (blunt injury, penetrating injury, and falls) and injury severity. We hypothesized that differences in likelihood of admission might be explained by these factors. Finally, we calculated the risks of death associated with admission to a nontrauma center vs a trauma center.
Our study was a retrospective analysis of adults (age, >18 years) hospitalized for trauma at acute care hospitals in California between January 1, 1999, and December 31, 2008. We combined and analyzed data from 3 separate data sets focusing on the state of California. The California Office of Statewide Planning and Development (OSHPD) patient discharge database contains discharge records from all nonfederal, licensed California hospitals, which are mandated to report all hospitalizations. Demographic variables include patient age, sex, zip code, and county of residence, International Classification of Diseases Ninth Revision (ICD-9) codes for principal diagnosis and up to 24 other diagnoses, as well as disposition variables.16
We identified which hospitals in our OSHPD patient-level data were level I or II trauma centers from the California Department of Health trauma center database.17 Because trauma center designation could change from one year to another, we assigned trauma center designation according to the patient's date of admission and hospital identifiers. We then linked the 2000 US census zip-code level income information to our study data set.
We selected adult patients admitted for traumatic injury using previously published criteria, defined by the principal or secondary diagnosis ICD-9 codes of 800-904.9, 910-929.9, and 950-959.9.18 To focus on acute trauma, we used an algorithm from previously published work19- 21 and excluded patients without external causes of injury codes (E codes), with scheduled admissions, or admitted for late effects of injury. Patients with ICD-9 codes for drowning, bites and stings, overexertion, poisonings or suffocation and patients whose sole ICD-9 codes indicative of injury were minor injuries (eg, sprains and strains, contusion with intact skin surface, or foreign body) were also excluded. Patients in California who have burns are often cared for at regional burn centers, so we also excluded patients whose principal diagnosis indicated a burn. Finally, because hip fractures are extremely common among the elderly and may not require all trauma center services, we excluded patients with closed hip fractures using prespecified ICD-9 codes.22- 23 This accounted for 97 006 visits, or 18.4% of what would have been our sample (n = 527 087). Excluding these closed hip fractures reduces the likelihood of exaggerated findings of bias against the provision of trauma center care for elderly patients given the large percentage of elderly patients with hip fractures (41.5% in what would have been our sample, compared with 4.0% of nonelderly patients) and is consistent with studies of injured elderly persons24 who could receive adequate care in nontrauma centers.
We used 2 separate predictors to represent elderly patients in our study because there are few clear definitions of “elderly” in relationship to trauma. The first was a dichotomous variable using 65 years of age as a cutoff.25- 26 We also constructed a categorical variable representing a series of standard age groupings27: 18-25 years, 26-45 years, 46-65 years, 66-85 years, and older than 85 years. This latter construction allowed us to examine whether associations were more or less powerful in standardized age groups.25- 29
Injury Severity Scores (ISSs) were calculated for patients using an International Classification of Diseases, Ninth Revision, Clinical Modification translation of the ISSs by Tri-Analytics (Bel Air, Maryland) using the “ignore unknowns” and “low severity” mapping options.30 If an ISS could not be computed, the patient was excluded. The ISS was then stratified into mild (ISS, 1-4), moderate (ISS, 5-15), and severe (ISS, >15) categories as suggested by MacKenzie et al.31 The mechanism of injury was determined from the principal ICD-9 code and from the E code, using the recommended framework for E-code groupings for injury mortality and morbidity from the Centers for Disease Control and Prevention.18
Because elderly patients likely have comorbidities that affect their overall prognosis, we calculated the Elixhauser Index, described and validated elsewhere in the literature,32 for each patient.
Whether a trauma center was located in a patient's county of residence was determined by comparing the patient's resident zip code to a list of zip codes for counties with a trauma center-designated hospital. We performed this analysis for each year because the designation of trauma center for a hospital can change over time. The geographic proximity of the patient's residence to a trauma center was calculated in miles (to convert miles to kilometers, multiply by 1.6) as the shortest distance between the centroid of the patient's zip code and the centroid of the nearest trauma center zip code, according to the method of Phibbs and Luft.33 Distance categories were chosen based on previous studies.19- 20
Insurance status was known from the medical record of each visit and was categorized into 5 categories: Medicare, Medicaid, private, uninsured or self-pay, and other. Household income was determined using the median household income for the patient's resident zip code, and 3 strata were created based on multiples of the 2000 federal poverty level or $18 850. Income strata were set according to previous literature20: (1) below 2 times the federal poverty level, (2) 2 to 3 times the federal poverty level, and (4) above 3 times the federal poverty level.
We first characterized our sample using univariable techniques and then compared age groups using bivariable methods, to compare the relative likelihood of receiving trauma center care as stratified by age. We then performed multivariable logistic regression analyses using an a priori selection model to assess the factors influencing the probability that patients received care at a trauma center. The demographic factors that were analyzed for inclusion in the regression model were age, sex, race/ethnicity, and insurance type. Injury characteristics in the model included injury severity (mild, moderate, or severe), type of injury (penetrating injury, blunt injury, or fall), and Elixhauser Index. System variables included proximity to trauma hospital (0-10 miles, 11-25 miles, 26-50 miles, or >50 miles), availability of trauma center in the county, and metropolitan statistical area status (urban or rural).
For our mortality predictions, we accounted for observable differences between our 2 groups of patients (those treated at trauma centers and those treated at nontrauma centers) using the previously described method of propensity scores or the inverse probability of treatment weighting approach.31,34 Briefly, the conditional probability of being admitted to a trauma center is calculated based on the known demographic and clinical characteristics of each patient. The data are weighted according to the reciprocal of this score to create a population in which the likelihood of admission to a trauma center is not confounded by other covariates and the effect of trauma center care reflects that of the original population.
We used SAS version 8.2 (SAS Institute, Cary, North Carolina). This protocol was approved by the institutional review boards for human research at Stanford University and the University of California, San Francisco.
Of the 430 081 patients admitted to California acute care hospitals for trauma-related diagnoses, 27% were older than 65 years. There were several notable differences in characteristics between the 2 groups (Table 1). Most trauma patients in the nonelderly group were men (63.0%); however, in the elderly group, this proportion was reversed: 65.2% of elderly trauma patients were women (P < .001). The nonelderly group had a higher proportion of trauma patients who were black than did the elderly group (7.0% vs 2.1%; P < .001). Similarly, there were fewer trauma patients who were white in the nonelderly group compared with the elderly group (32.8% vs 66.4%; P < .001).
When patients were stratified by injury severity, there were fewer severely injured elderly patients than nonelderly patients (3.9% vs 7.3%; P < .001) but slightly more moderately injured elderly patients than nonelderly patients (41.4% vs 39.3%; P < .001). More elderly patients had a higher number of comorbidities. By mechanism, the elderly patients had a disproportionate number of falls (Figure 1) compared with the nonelderly patients (81.5% vs 28.3; P < .001). More than 93% of injuries in those 85 years of age or older are due to falls, compared with 14% in those 18-25 years of age.
Percentage of trauma injuries by mechanism of injury, stratified by age group.
Distance to trauma center, metropolitan statistical area status, and availability of trauma center within county of residence were similar between groups. However, a much larger percentage of the elderly patients were admitted to nontrauma centers compared with the nonelderly patients (71.1% vs 36.8; P < .001). A higher proportion of the elderly patients than nonelderly patients (3.9% vs 2.1%; P < .001) also died in the hospital after admission for their injuries.
We first constructed bivariate models of the likelihood of an elderly patient (defined first in the binary definition as being older than 65 years of age) vs a nonelderly patient (defined first in the binary definition as being between 18 and 65 years of age) being admitted to a trauma center. With this method, the odds ratio (OR) of an injured elderly patient compared with a nonelderly patient receiving care in a trauma center was 0.23 (95% confidence interval [CI], 0.18-0.29).
The results of our multivariable regression model of older patients as defined by the binary categorization of older or younger than 65 years demonstrated that, even after adjusting for available patient-level risk factors and access to a trauma center, patients older than 65 years still had a dramatically decreased likelihood (OR, 0.53; 95% CI, 0.45-0.63) of being admitted to a trauma center compared with patients 65 years of age or younger.
We then examined the proportion of injured patients admitted to a trauma center by age categories (Figure 2) and saw a linear relationship between age and referral likelihood. Even after controlling for predictors, increasing age showed a profound association with lower likelihood of trauma center referral in the multivariable regression (Table 2). The linear trend of the inverse association of age and likelihood of admission to trauma center persisted: patients aged 46-65 years had a 0.57 odds (95% CI, 0.54-0.60) of receiving trauma care compared with the referent population aged 18-25 years; patients 66-85 years of age had an OR of 0.35 (95% CI, 0.30-0.41); and patients 85 years or older had an OR of 0.30 (95% CI, 0.25-0.36).
Percentage of injured patients admitted to trauma center (TC) by age group, stratified by year.
As a secondary aim, we sought to determine whether these differences in likelihood of admission varied across injury type (blunt injury, penetrating injury, and falls) as well as injury severity, because certain mechanisms or certain levels of severity could trigger more alarm in both the prehospital and in-hospital setting. We found, contrary to our hypothesis, that this pattern of decreased likelihood of receiving care in a trauma center for older patients was unchanged even in the adjusted multivariable analyses when stratified by injury type and severity (Figures 3 and 4).
Likelihood of admission to trauma center (TC) by mechanism of injury, stratified by age group. Likelihood of admission to TC for blunt trauma (n = 167 652) (A), penetrating trauma (n = 77 843) (B), and fall (n = 167 652) (C).
Likelihood of admission to trauma center (TC) by injury severity, stratified by age group. Likelihood of admission to TC for mild trauma (n = 231 200) (A), moderate trauma (n = 171 504) (B), and severe trauma (n = 27 378) (C).
Given that protocols of trauma transport may differ across emergency medical services that are county-based in California, we performed additional modeling to include clustering by county. Our results were robust to clustering by county as well as over time. We also performed these analyses to include all levels of trauma centers (ie, levels I-IV) and did not find that this significantly altered our results; older patients did not only have a decreased likelihood of receiving care in level I and II trauma centers. Models to account for interaction of age (considering both binary and ordinal definitions) with injury severity (considering continuous and categorical definitions) did not change our results.
Additionally, we ran stratified analyses of these models by age group to determine whether certain factors were more or less predictive of admission to trauma center. For patients 18-65 years of age, those living within a county with a trauma center had a 6.4 odds (95% CI, 3.1-13.3) of being admitted to a trauma center. For patients older than 65 year, those living within a county with a trauma center had 3.9 odds (95% CI, 2.0-7.6) of being admitted to a trauma center. In other words, the availability of a trauma center in one's county was less beneficial for elderly patients than nonelderly patients.
Elderly patients in California have a lower likelihood of receiving trauma center care, after controlling for injury severity, insurance, income, proximity to a trauma center, and availability of a trauma center within the patient's county. We show not simply a disparity for the oldest patients but rather a proportional decrease in likelihood of receiving trauma center care according to age. This disparity of who receives trauma care persists across a variety of trauma types, from blunt injury, penetrating injury, to falls. Furthermore, the decreased likelihood of trauma care for the elderly is not only present for mild trauma but also, when examined separately, for moderate and severe trauma. Finally, we show that these disparities have continued over time.
Our study confirms findings from smaller studies suggesting that elderly patients are less likely to be admitted to a trauma center for their injuries9,13 and extends these findings in 2 significant ways. First, we account for structural factors not usually considered in other studies, including regional availability of trauma centers and proximity to trauma center, which are both known to affect the likelihood of admission to a trauma center. Second, our study provides a truly population-level view of the growing public health problem posed by disparities in access to trauma care. Our data suggest that elderly trauma patients have a lower likelihood of being taken to a trauma center (suggesting prehospital bias)35 or of being transferred to a trauma center once taken to a nontrauma center (in-hospital bias).36
Although we have fewer data to discern potential mechanisms underlying referral biases, we do know that the proportion of elderly patients experiencing isolated, low-severity injury has been increasing20 and that there is agreement that these patients benefit less from trauma care.7 It is important to note, however, that our methods excluded patients with hip and femur fractures, and that our findings were consistent across all levels of injury severity. In addition, these explanations would not sufficiently explain the consistency of our findings even when evaluating only severely injured patients.
An additional, more insidious, possibility of why elderly patients may have a lower likelihood of receiving care in a trauma care is the conscious or subconscious undertriage of care in the elderly population due to less valuation of years of life lost. There is a very real concern of age-based referral bias and age-based rationing of care, which has been documented in areas other than trauma.37- 39 Because elderly patients have higher mortality rates from trauma compared with younger patients, it is possible that, as posited in other diseases such as myocardial infarction,37 more aggressive intervention could theoretically be more beneficial for an older population, as suggested by our findings of decreased in-hospital mortality for hospitalized elderly patients with severe trauma compared with nonelderly patients.
A natural extension of this work would seek to clarify the precise reasons underlying the disparity found in our study. Elderly patients may have unclear or benign initial clinical presentations, leading to undertriage,40- 42 or a lack of compliance with trauma triage rules may be at play.12 The American College of Surgeons and the Centers for Disease Control and Prevention have recommended that trauma patients older than 55 years should automatically be taken to a designated trauma center regardless of severity of injury.43- 44 Other studies40,45 have suggested that trauma patients older than 60 years or 70 years should be considered as fulfilling a criterion for automatic trauma activation. However, such broad inclusion criteria would have significant implications for the delivery of care and possible overtriage,46 given the limited resources available for trauma care. Currently, there is much debate in the emergency and trauma community about inappropriate trauma “activations” (which alert the hospital to the need for trauma resources such as the surgical and anesthesia team) from the field during the prehospital phase,47- 49 or while in hospital,50 and unnecessary transfers that deplete the hospital of needed resources and other patients.51 In fact, studies of centers that do have age as a criterion for automatic trauma activation demonstrate that such regulation can result in unnecessary and costly initiations of a multidisciplinary response.52
Ironically, overtriage has also been suggested to be detrimental to the patient's health.53- 54 From previous studies of pediatric patients in which age has been considered a criterion for an automatic trauma response,55- 58 it has been found that age (specifically young age) may not be an appropriate sole criterion for requiring a higher level of trauma care. There is, however, some literature suggesting a tiered approach to trauma center activation once a patient has reached the emergency department for a range of patient characteristics.59- 60 Overall, triage protocols and their effects on undertriage and overtriage must be evaluated for all age ranges to study the effects on the primary patient as well as other patients within the hospital and system that account for appropriate resource utilization.61- 63
Our study has a number of limitations. We captured hospital admissions, which represent only a portion of the patients seen by emergency medical teams. Because patients must be admitted to the hospital to be represented in this database, our data do not include trauma patients discharged from the emergency department who generally have low injury severity and mortality rates64 or patients who died before admission and who, by definition, have high mortality and likely high injury severity rates.
Second, our adjustment for trauma severity was based on administrative data; a full range of clinical variables cannot be taken into account. Although the ISS has been validated for trauma patients in general (predicting both mortality and hospital stay) and correlates with tools using clinical data, it is important to note that ISS is not linear and has not been validated in the subset of geriatric patients.65- 66
Finally, we use trauma center designations provided by the OSHPD, which does not conform to what many trauma surgeons consider the gold standard of categorization as stated by the American College of Surgeons Committee on Trauma.67 We were unable to use these categorizations because verification of these centers by the American College of Surgeons began after the period of our study. We do not, however, expect systematically differential reporting or categorization, so it is highly unlikely that this limitation affects our results.
In our analysis of elderly trauma patients admitted to acute care hospitals, we have demonstrated that there is a significant correlation between age and the likelihood of admission to a trauma center such that increasing age ominously decreases the likelihood of trauma center admission, even among those with moderate and severe injury. This disparity in the treatment of elderly trauma patients is alarming, because demographic trends forecast that the proportion of injured elderly patients will only increase in the years to come. The elderly patients who are at risk for higher mortality from their injuries may in fact be a subpopulation that is most likely to benefit from more aggressive care. Controlled studies of prehospital triage of elderly patients as well as provider beliefs about trauma triage in elderly patients are important for designing interventions to better target available resources such as trauma centers for injured elderly patients.
Correspondence: Renee Y. Hsia, MD, MSc, Department of Emergency Medicine, University of California, San Francisco General Hospital, 1001 Potrero Ave, 1E21, San Francisco, CA 94110 (email@example.com).
Accepted for Publication: October 28, 2010.
Published Online: January 17, 2011. doi:10.1001/archsurg.2010.311
Author Contributions: Dr Hsia had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Hsia, Wang, Wise, and Auerbach. Acquisition of data: Wang, Wise, and Auerbach. Analysis and interpretation of data: Hsia, Wang, Saynina, Wise, Pérez-Stable, and Auerbach. Drafting of the manuscript: Hsia and Auerbach. Critical revision of the manuscript for important intellectual content: Wang, Saynina, Wise, Pérez-Stable, and Auerbach. Statistical analysis: Saynina and Auerbach. Obtained funding: Hsia and Pérez-Stable. Administrative, technical, and material support: Hsia, Wang, Wise, Pérez-Stable, and Auerbach. Study supervision: Pérez-Stable and Auerbach.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant P30-AG15272 from the Resource Centers for Minority Aging Research program funded by the National Institute on Aging (Drs Hsia and Pérez-Stable), by NIH/NCRR/OD UCSF-CTSI grant KL2 RR024130 (Dr Hsia) and NIH/NICHD grant K23 HD051595 (Dr Wang) from the National Institutes of Health, and by the Robert Wood Johnson Foundation Physician Faculty Scholars (Dr Hsia).
Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the opinions of any of the funding agencies.
Additional Contributions: We thank David Wang, MD, for his help with this article.
Country-Specific Mortality and Growth Failure in Infancy and Yound Children and
Association With Material Stature
Use interactive graphics and maps to view and sort country-specific infant and early
dhildhood mortality and growth failure data and their association with maternal
Thank you for submitting a comment on this article. It will be reviewed by JAMA Surgery editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Register and get free email Table of Contents alerts, saved searches, PowerPoint downloads, CME quizzes, and more
Subscribe for full-text access to content from 1998 forward and a host of useful features
Activate your current subscription (AMA members and current subscribers)
Purchase Online Access to this article for 24 hours
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 4
Customize your page view by dragging & repositioning the boxes below.
and access these and other features:
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.