Printer Friendly

Development of a middle-age and geriatric trauma mortality risk score: a tool to guide palliative care consultations.

Trauma is the fifth leading cause of death in geriatric adults, and the geriatric patient represent 12% of the general trauma population. (1,2) By 2050, geriatric people will account for over 20% of the population and at least 40% of the general trauma population. (2,3) Current literature suggests that older trauma patients have increased rates of morbidity and mortality when compared to younger patients with similar injuries. Despite the high risk of inpatient mortality in these older patients, early intervention in patient care with goal-directed services such as palliative medicine services is lacking. (4,5) One limitation of the current care pathway in organized trauma centers is the difficulty in identifying geriatric and middle aged patients at high risk for inpatient mortality early on in the hospital course. The goal would be to identify such patients prior to the initiation of any resource intensive endeavors. Realistically, high risk patients within this cohort would benefit from early discussion centered upon goals of care.

Currently, injury severity indices are used to characterize mortality risk within the trauma population. These tools have been extensively studied within the young adult population. (6,7) However, their usefulness in the middle-age and geriatric population is not as clearly defined. Current research suggests many trauma protocols fail to address the underlying physiologic changes and comorbid pre-existing conditions prevalent within this population. (8)

Previous studies have also failed to recognize the importance of the mechanism of injury within the older trauma population. These studies have combined both low-energy and high-energy older trauma patients, which may explain the poor predictive capacity demonstrated by current injury severity indices. (8-10) Patients at risk for inpatient mortality following low-energy injury may represent a very different population than those at high risk following high-energy injury.

Accordingly, we sought to develop a novel score to calculate risk of inpatient mortality in low-energy and high-energy geriatric and middle age patients (e.g., Score for Trauma Triage in Geriatric and Middle-Age, STTGMA). Within this mortality risk assessment we stratified patients based upon mechanism of initial injury (e.g., [STTGMA.sub.LE] and [STTMGA.sub.HE]). We defined "middle-aged" as between 55 to 64 years of age and "geriatric" as 65 years of age or older for the purposes of triage decisions due to evidence that differences in outcome rise sharply with increasing age. (11-17) This phenomenon may be explained by the increase in prevalence of comorbidity and the observed decrease in physiologic reserve among patients of these two groups.

The study group aimed to develop this risk model based upon the experience of one level 1 trauma center with a high volume of geriatric and middle-aged trauma patients. The purpose of this study was to create a predictive tool able to augment clinical decision making early on in the hospital course. The hope is that early high-fidelity patient identification will result in opportunities to improve patient care, provide improved prognostic ability, and limit hospital resource expenditure as measured by cost. The investigators devised a theoretical model of expected reduction in direct hospital costs when patients determined to be at high risk of inpatient mortality were evaluated by palliative medicine early in the hospitalization. Using published guidelines on expected reductions in direct hospital costs, the investigators aimed to demonstrate the ability of the novel mortality risk score to identify opportunities for early intervention and quality improvement.

Patients and Methods

We conducted a retrospective review of data entered into the Trauma Registry for all trauma patients aged 55 years and older who presented to our Level I trauma center during a 4-year period (January 1, 2008, through December 31, 2011). Additionally, the Trauma and Injury Severity Score (TRISS) uses an age index in its calculation with a dichotomous variable of < 55 or [greater than or equal to] 55 years old, therefore we felt 55 to be an acceptable age cutoff for this study. A total of 5,710 patients were divided into low-energy and high-energy cohorts. Low-energy mechanisms included all ground-level falls (including isolated hip fractures) [(International Classification of Diseases, 9th Edition (ICD-9) (e) codes e880-e884)]. High-energy mechanisms included all motor vehicle accidents, motorcycle crashes, pedestrians struck by motor vehicles, and falls from height (ICD-9 e-codes e881, e882, e884.1). A total of 2,122 patients were excluded from the initial analysis: 2,093 patients with incomplete data [including 957 with missing Abbreviated Injury Score (AIS)], and 29 patients who were dead on arrival to the emergency department. This left 3,588 patients available for analysis: 2,387 low-energy and 1,201 high-energy geriatric and middle-age trauma patients. For the total sample, Pearson's chi-square test was used to compare gender, mechanism of injury, and inpatient mortality rates. Student's t-test was used to compare age and mortality rates.

In Phase 1, backward stepwise logistic regression analysis was used to develop the [STTGMA.sub.LE] and [STTMGA.sub.HE]. We used an initial significance threshold of p < 0.20 for inclusion in the model while the final model included only independent predictors of in-hospital mortality with a significance level of p < 0.05. The four host factors we studied were age, pre-existing conditions, emergency department (ED) vital signs, and anatomic injuries. Age was analyzed as a continuous variable and anatomic injuries were analyzed as dichotomous outcomes (presence or absence) and as ordinal variables using the Abbreviated Injury Severity (AIS) scale for face, head and neck, chest, abdomen, extremities, and soft-tissues. This scale orders anatomic injuries from 0 to 6 in order of increasing injury severity (0 = no injury, 6 = unsurvivable injury). Emergency department vital signs included heart rate (HR), systolic blood pressure (SBP), respiratory rate (RR), and Glasgow Coma Scale (GCS). All four vital signs were analyzed as continuous variables and SBP, RR, and GCS were also evaluated as ordinal variables based on coded Revised Trauma Score (RTS) values. (18) The different iterative versions of the models we tested are detailed in Tables 1 and 2. Discrimination and calibration of the models was performed using AUROC analysis and the Hosmer-Lemeshow (H-L) statistic, respectively. (19) We selected the model with the highest AUROC and best calibration as the [STTGMA.sub.LE] and [STTMGA.sub.HE].

In Phase 2, we used 4 years (2008 to 2011) of the National Trauma Databank Research Dataset (NTDB RDS) to validate our newly developed trauma scores. We identified 59,965 low-energy and 97,034 high-energy geriatric and middle-age trauma patients with complete data (all variables included in our newly developed score) that formed the basis of the validation cohort. We used Students t-test to compare differences between the variables that comprised the [STTGMA.sub.LE] and [STTMGA.sub.HE] in the NTDB and Carolinas cohort. We compared the AUROC of [STTGMA.sub.LE] and [STTMGA.sub.HE] against the Trauma and Injury Severity Score (TRISS), the most commonly used trauma mortality risk tool, in the NTDB cohort using the method devised by Hanley and McNeil (19) to determine if there was a difference in predictive capacity for in-hospital mortality.

Finally, in phase 3 we created risk stratification groups based on [STTGMA.sub.LE] and [STTMGA.sub.HE] inpatient mortality risk defined as STTGMA 0 (0% to 0.9%), 1 (1.0% to 2.9%), 2 (3.0% to 9.9%), and 3 (10.0% to 100.0%) to indicate low, intermediate, high, and very-high risk patients, respectively. Relative risk of mortality was calculated for each tier with STTGMA 0 used as the control group. We then used guidelines published by the Palliative Care Leadership Centers' Outcomes Group regarding cost savings associated with palliative care consultation programs to calculate the expected reduction in net and direct hospital costs if all STTGMA 2 and 3 patients had received a palliative care consultation on admission. Note that net hospital costs is a reflection of direct and indirect hospital costs. We used a reduction of net hospital costs by $2,642 per admission for patients discharged alive and $6,896 per admission for patients who died during hospitalization. (5) We used a reduction of direct hospital costs by $1,696 per admission for patients discharged alive and $4,908 per admission for patients who died during hospitalization. (5) Results were reported as cost savings (in dollars) per year for low-energy and high-energy cohorts.

Results

A total of 3,588 patients were available for comparison of low-energy and high-energy geriatric trauma patients. Low-energy patient were 9.1 years older than high-energy patients (75.8 [+ or -] 11.0 years vs. 66.7 [+ or -] 9.2 years, p < 0.01). Two-thirds of low-energy patients were female compared to only onethird of high-energy patients (p < 0.01). There were 2,387 (66.5 %, 2387/3588) patients who sustained a low-energy fall from standing height with a resultant inpatient mortality rate of 7.9% compared to 1,201 (33.5%, 1201/3588) patients who sustained high-energy blunt trauma resulting in a 7.0% mortality rate (p = 0.21).

Phase 1

Score for Trauma Triage in Middle-Age and Geriatric Trauma Patients--Low Energy: Logistic Regression Analysis to Formulate Model to Predict In-Hospital Mortality

Our logistic regression analysis using variables from the four core host factors (age, PECs, ED vital signs, and anatomic injuries) yielded the following equation with the highest AUROC (0.89) and the best calibration (H-L = 8.34) for predicting the probability of in-hospital mortality (see Table 1 for different models tested):

Probability of In-Hospital Mortality = 1 / (1 [+ or -] exp((-3.41 [+ or -] 0.05 * AGE [+ or -] 0.25 * CCI-0.33*GCS [+ or -] 0.51 * AIS_ HN [+ or -] 0.42 * AIS_CHS)))*

Where: Age (continuous variable); CCI = Charlson Comorbidity Index (ordinal variable); GCS = Glasgow Coma Scale (continuous variable); AIS_HN = Head and Neck Abbreviated Injury Score (ordinal variable); and AIS_CHS = Chest Abbreviated Injury Score (ordinal variable).

Analysis of the Individual Independent Predictors of In-Hospital Mortality Used in the [STTGMA.sub.LE]

Age

The mean age of low-energy patients was 75.9 [+ or -] 11.1 years. Each 1 year increase in age was associated with a 5% increase in relative risk of mortality (OR 1.05 CI: 1.03-1.07). We found a significantly higher mortality rate for patients aged 75 years or older versus those less than 75 years of age (9.3% vs. 5.9%, p < 0.01).

CCI

The mean CCI was 0.96 [+ or -] 1.4. Each 1 point increase in CCI was associated with a 28% increase risk of mortality (OR 1.28 CI: 1.14-1.44). There was a significant difference in mortality rates for patients with a CCI of 4 or more versus those with less than 4 (12.1% vs. 7.5%, p = 0.04).

ED Vitals (GCS)

The mean GCS was 14.1 [+ or -] 2.6. Each 1 point decrease in GCS was associated with a 28% increased risk of mortality (OR 0.72 CI: 0.69-0.76). There was a significant difference in mortality rates for patients with a GCS 14 or less versus those with 15 (23.6% vs. 2.6%, p < 0.01).

Anatomic Injuries (AIS_HN and AIS_CHS)

The mean AIS_HN was 1.6 [+ or -] 1.9. Each 1 point increase in AIS_HN was associated with a 67% increase risk of mortality (OR 1.67 CI: 1.49-1.87). The mean AIS_CHS was 0.2 [+ or -] 0.7. Each 1 point increase in AIS_CHS was associated with a 52% increase risk of mortality (OR 1.52 CI: 1.19-1.92).

Score for Trauma Triage in Middle-Age and Geriatric Trauma Patients--High Energy: Logistic Regression Analysis to Formulate Model to Predict In-Hospital Mortality

Our logistic regression analysis using variables from the four core host factors (age, PECs, ED vital signs, and anatomic injuries) yielded the following equation with the highest AUROC (0.96) and the best calibration (H-L = 4.28) for predicting the probability of in-hospital mortality (see Table 2 for different models tested):

Probability of In-Hospital Mortality = 1 / (1 [+ or -] exp(-(8.69 [+ or -] 0.11*AGE-0.37* GCS [+ or -] 0.57*AIS_HN [+ or -] 0.41*AIS_ CHS [+ or -] 0.46*AIS_EXT)))

Where: Age (continuous variable); GCS = Glasgow Coma Scale (continuous variable); AIS_HN = Head/Neck Abbreviated Injury Score (ordinal variable); AIS_CHS = Chest Abbreviated Injury Score (ordinal variable); and AIS__EXT = Extremity Abbreviated Injury Scale (ordinal variable).

Analysis of the Individual Independent Predictors of In-Hospital Mortality Used in the [STTGMA.sub.HE] Age

The mean age of high-energy cohort was 66.7 [+ or -] 9.3 years. Each 1 year increase in age was associated with a 1.1% increase in relative risk of mortality (OR 1.08 CI: 1.03-1.17).

ED Vitals (GCS)

The mean GCS was 14.1 [+ or -] 2.8. Each 1 point decrease in GCS was associated with a 31% increase risk of mortality (OR 0.69 CI: 0.64-0.75). There was a significant difference in mortality rates for patients with a GCS less than 14 versus those with 15 (20.4% vs. 1.0%, p < 0.01).

Anatomic Injuries (AIS_HN, AIS_CHS, & AIS_EXT)

The mean AIS_HN was 0.8 [+ or -] 1.5. Each 1 point increase in AIS_HN was associated with a 77% increase risk of mortality (OR 1.77 CI: 1.41-2.21). The mean AIS_CHS was 1.0 [+ or -] 1.4. Each 1 point increase in AIS_CHS was associated with a 51% increase risk of mortality (OR 1.51 CI: 1.20-1.90). The mean AIS_EXT was 0.9 [+ or -] 1.2. Each 1 point increase in AIS_EXT was associated with a 59% increase risk of mortality (OR: 1.14-2.21).

Table 3 shows a side-by-side comparison of variables found to be independent predictors of mortality in the high-energy and low-energy cohorts.

Phase 2

With regards to low-energy mechanisms of injury, the NTDB cohort (N = 59,965) and Carolinas cohort (N = 2,387) demonstrated no differences in AIS_HN scores. Statistically significant differences between age, GCS, AIS_CHS, and CCI were identified between cohorts (p < 0.01), (Table 4).

With regards to high-energy mechanisms of injury, the NTDB cohort (N = 97,034) and Carolinas cohort (N = 1,201) demonstrated no differences in GCS scores. Statistically significant differences between age, AIS_CHS, AIS-HN, and AIS-EXT were identified between cohorts (p < 0.01), (Table 4).

The [STTGMA.sub.LE] was noted to have significantly greater predictive capacity compared to TRISS in the NTDB cohort (AUROC 0.83 vs. 0.80, p < 0.01). Similarly, the [STTGMA.sub.HE] was noted to have significantly greater predictive capacity compared to TRISS in the NTDB cohort (AUROC 0.86 vs. 0.85, p < 0.01), (Table 4).

Phase 3

In the low-energy cohort, 28.2% (674/2,387) patients and 16.7% (399/2,387) of patients were risk stratified into [STTGMA.sub.LE] 2 and 3, respectively. The relative risk of mortality for [STTGMA.sub.LE] 2 and 3 was 28.2 (3.9-204.6 95% CI, p < 0.01) and 173.0 (24.3-1232.0 95% CI, p < 0.01), respectively. In our theoretical model, a palliative care consultation on admission for all patients in [STTGMA.sub.LE] 2 and 3 would have resulted in $886,321 reduction in net hospital costs per year whereas a consult on all admission in STTGMELE 3 only would have resulted in $402,858 reduction in net hospital costs per year. In the high-energy cohort, 9.6% (115/1201) patients and 8.7% (104/1201) patients were risk stratified into [STTGMA.sub.HE] 2 and 3, respectively. The relative risk of mortality for STTGMA 2 and 3 was 67.5 (8.7-522.2, 95% CI, p < 0.01) and 291.0 (40.4-2095.0, 95% CI, p < 0.01), respectively. In our theoretical model, a palliative care consultation on admission for all patients in [STTGMA.sub.HE] 2 and 3 would have resulted in $196,761 reduction in net hospital costs per year whereas a consult on all admissions in [STTGMA.sub.HE] 3 only would have resulted in $110,169 reduction in net hospital costs per year. The overall projected reduction in net hospital costs was between $1,083,082 ([STTGMA.sub.LE[+ or -]HE] 2 and 3) and $513,027 ([STTGMA.sub.LE[+ or -]HE] 3 only) per year. Table 5 shows a complete breakdown of patient distribution, relative risk of mortality for [STGGMA.sub.HE] and [STTGMA.sub.LE] 0, 1, 2, and 3, and net reduction in hospital costs associated with palliative care consultation.

Discussion

The [STTGMA.sub.LE] and [STTGMA.sub.HE] tools are novel middle-age and geriatric-specific inpatient mortality risk calculators that account for mechanism of injury: a factor previously left unaddressed. The calculation of these risk scores requires information readily available upon initial patient evaluation. The validation of the risk tool within the National Trauma Databank demonstrated the tool's clinical utility within a large and heterogeneous population beyond the experience of a single level 1 trauma center. The ability of our aged patient-specific risk tool to outperform the widely used Trauma and Injury Severity Score (TRISS) index to predict inpatient mortality within a nationwide patient cohort confirms its utility.

Both TRISS and the novel score developed here are driven in large part by head injury and observed mental status on presentation; given that 3/5 and 2/5 core variables used to calculate the scores, respectively, deal with this issue in some form. While TRISS also gives importance to head injury and mental status, it accounts for only 2/10 variables required to predict mortality. The role of geriatric head injury has been extensively studied. Mortality rates for geriatric patients with GCS of 7 or less have been reported to be as high 74%. (21) In the low-energy cohort, a GCS less than 15 was associated with a mortality rate of 23.6%; an odds ratio of 9.1 when compared to those with a GCS of 15. Similarly, in the high-energy cohort, a GCS score less than 15 was associated with a mortality rate of 20.4%; an odds ratio of 20.4 when compared to those with a GCS of 15 on presentation.

The significance of pre-existing conditions in contributing to the risk-assessment profile for geriatric trauma patients has been mixed in the literature. (25) In the current study, we found that the Charlson Comorbidity Index score was an independent predictor of in-hospital mortality in the low energy cohort only and that each one point increase in CCI was associated with a 28% increase risk of mortality. Pre-existing conditions were not found to be predictive of in-hospital mortality in the high-energy cohort. This likely demonstrates that the underlying health of the patient plays a greater role in low energy trauma, and the nature of the injury itself is more important in high energy injuries. As such, pre-existing conditions are included in the low-energy score but not in the high-energy score. TRISS does not include comorbidities. This supports our finding that pre-existing conditions do not play a significant factor in predicting in-hospital mortality in the high-energy cohort given than both the [STTGMA.sub.HE] and TRISS were derived from high-energy mechanism cohorts. Several large retrospective studies have evaluated the role of pre-existing conditions in older trauma patients and found conflicting evidence regarding their predictive capacity with advancing age greater than 55. (25,26) Findings from previous studies suggest that comorbid conditions may be predictive of in-hospital complications but not overall in-hospital mortality. (27-28) Nearly all studies evaluating the role of pre-existing conditions in geriatric trauma patients have combined low-energy and high-energy patient cohorts. In contrast, our group has previously evaluated specific co-morbidities derived from the Charlson Comorbidity Index in 2,940 low-energy geriatric and middle-age trauma patients and found that moderate to severe renal disease, diabetes with end-organ damage, and leukemia and lymphoma were independent predictors of in-hospital mortality. (10)

It is well documented that palliative care consultations are underutilized in the surgical setting. In a 2014 national survey of 362 trauma surgeons affiliated with the Eastern Association for the Surgery of Trauma (EAST), nearly half (46%) felt that palliative care consultations were underutilized at their Level 1 trauma center. (4) In this same study, the investigators found that the most commonly cited benefits of these consultations were assistance with end-of-life issues (72.9%), decrease in futility (60.9%), and communication of prognosis (49.4%). (20) The Palliative Care Leadership Centers' Outcomes Group has shown that significant reduction in hospital costs can be achieved with dedicated palliative care consultation teams. (5) The reasons for the reduction in costs is attributed primarily to the "fundamental shift from the usual hospital pathway" to one that is more in-line with end-of-life goals leading to a legitimization of the discontinuation of certain treatments.

In Phase 3 of our study, we demonstrated the role our new scores could play in quality improvement as they pertain to risk stratifying patients based upon their inpatient mortality risk. In our theoretical cost-saving model based upon the Palliative Care Leadership Centers' Outcomes Group published data, we projected an annual reduction in net hospital costs between $513,027 to $1,083,082 utilizing early palliative medicine consultation. While these estimations are theoretical, they do highlight the need for further prospective study. Goals for future study would be demonstration of real-world cost savings through early identification of high risk-resource intensive middle-age and geriatric trauma patients.

Our study has two key limitations. The retrospective nature of this study limits the analysis to the data which was available within the local and national trauma registries. This group of investigators feels that any lapses in data capture were adequately offset by the large sample size of the National Trauma Databank. The cost saving's model put forth by this group was based upon the data reported by one group of palliative medicine physicians. These projections do, however, demonstrate the need for future prospective study regarding early intervention in the care of high risk patients. Interestingly, the cost saving's estimates used within this study were based upon interventions at hospitalization days 7, 10, and 15. It is the belief of this group that greater resource savings would be achieved with our risk tool, as it can determine risk at the time of initial patient evaluation. This would allow for intervention within the first 48 hours of hospitalization and, thus, have a greater impact on the trajectory of patient care. The hope of this group and all caregivers to this population is that patient centered goals of care could be established prior to the initiation of much unwanted invasive intervention.

In conclusion, we have developed and validated a mortality risk assessment tool, the [STTGMA.sub.LE] and [STTMGA.sub.HE], that is specific to the middle-age and geriatric trauma populations and accounts for unique role that mechanism of injury plays in this population. Given the advent of the age of the electronic medical record system, with its immense potential to provide patient information at the onset of care, many institutions already collect this data upon patient arrival. Real-time implementation of data collection has already begun at the investigator's home institution without difficulty. The goal of this tool is to soon serve as a real-time clinical support tool in the areas of improved patient care, patient experience, resource allocation, prognostic ability, and clinical decision making.

Disclosure Statement

None of the authors have a financial or proprietary interest in the subject matter or materials discussed, including, but not limited to, employment, consultancies, stock ownership, honoraria, and paid expert testimony.

References

(1.) McMahon DJ, Schwab CW, Kauder D. Comorbidity and the elderly trauma patient. World Surg. 1996; 20:1113-9; discussion 9-20.

(2.) Koval KJ, Meek R, Schemitsch E, et al. An AOA critical issue. Geriatric trauma: young ideas. J Bone Joint Surg Am. 2003; 85-A(7):1380-8.

(3.) Campbell JW, Degolia PA, Fallon WF, et al. In harm's way: Moving the older trauma patient toward a better outcome. Geriatrics. 2009; 64(1):8-13.

(4.) Karlekar M, Collier B, Parish A, et al. Utilization and determinants of palliative care in the trauma intensive care unit: results of a national survey. Palliat Med. 2014; 28(8):1062-8.

(5.) Morrison RS, Penrod JD, Cassel JB, et al. Cost savings associated with US hospital palliative care consultation programs. Arch Intern Med. 2008; 168(16):1783-90.

(6.) Baker SP, O'Neill B, Haddon W Jr, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974; 14(3):187-96.

(7.) Osler T, Baker SP, Long W. A modification of the injury severity score that both improves accuracy and simplifies scoring. J Trauma. 1997(6); 43:922-5; discussion 5-6.

(8.) Jacobs DG, Plaisier BR, Barie PS, et al. Practice management guidelines for geriatric trauma: the EAST Practice Management Guidelines Work Group. J Trauma. 2003; 54(2):391-416.

(9.) Konda SK, Lack WD, Seymour RB, Karunakar MA. Mechanism of injury differentiates risk factors for mortality in geriatric trauma patients. J Orthop Trauma. 2015; 29(7):331-6.

(10.) Konda S, Seymour R, Karunakar M. Predictors of mortality in geriatric trauama: an opportunity for triage. Presented at the Annual Scientific Meeting of the American Geriatric Society, Orlando, Florida, May 15-18, 2014.

(11.) Grossman M, Scaff DW, Miller D, et al. Functional outcomes in octogenarian trauma. J Trauma. 2003; 55(1):26-32.

(12.) Hannan EL, Waller CH, Farrell LS, et al. Elderly trauma inpatients in New York State: 1994-1998. J Trauma. 2004; 56(6):1297-304.

(13.) Sharma OP, Oswanski MF, Sharma V, et al. An appraisal of trauma in the elderly. Am Surg. 2007; 73(4):354-8.

(14.) Gowing R, Jain MK. Injury patterns and outcomes associated with elderly trauma victims in Kingston, Ontario. Can J Surg. 2007; 50(6):437-44.

(15.) Giannoudis PV, Harwood PJ, Court-Brown C, et al. Severe and multiple trauma in older patients; incidence and mortality. Injury. 2009; 40(4):362-7.

(16.) Taylor MD, Tracy JK, Meyer W, et al. Trauma in the elderly: intensive care unit resource use and outcome. J Trauma. 2002; 53(3):407-14.

(17.) Aitken LM, Burmeister E, Lang J, et al. Characteristics and outcomes of injured older adults after hospital admission. J Am Geriatr Soc. 2010; 58(3):442-9.

(18.) Champion HR, Sacco WJ, Copes WS, et al. A revision of the trauma score. J Trauma. 1989; 29:623-9.

(19.) Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982; 143(1):29-36.

(20.) Mosenthal AC, Murphy PA, Barker LK, et al. Changing the culture around end-of-life care in the trauma intensive care unit. J Trauma. 2008; 64(6):1587-93.

(21.) Howard MA 3rd, Gross AS, Dacey RG Jr, et al. Acute subdural hematomas: an age-dependent clinical entity. J Neurosurg. 1989; 71(6):858-63.

(22.) Sampson EL, Blanchard MR, Jones L, et al. Dementia in the acute hospital: prospective cohort study of prevalence and mortality. Br J Psychiatry. 2009; 195(1):61-6.

(23.) Benesch CG, McDaniel KD, Cox C, et al. End-stage Alzheimer's disease. Glasgow Coma Scale and the neurologic examination. Arch Neurol. 1993; 50(12):1309-15.

(24.) Sporer KA, Solares M, Durant EJ, et al. Accuracy of the initial diagnosis among patients with an acutely altered mental status. Emerg Med J. 2013; 30(3):243-6.

(25.) Morris JA Jr, MacKenzie EJ, Damiano AM, et al. Mortality in trauma patients: the interaction between host factors and severity. J Trauma. 1990; 30(12):1476-82.

(26.) Milzman DP, Boulanger BR, Rodriguez A, et al. Pre-existing disease in trauma patients: a predictor of fate independent of age and injury severity score. J Trauma. 1992; 32(2):236-43; discussion 43-4.

(27.) Sampalis JS, Nathanson R, Vaillancourt J, et al. Assessment of mortality in older trauma patients sustaining injuries from falls or motor vehicle collisions treated in regional level I trauma centers. Ann Surg. 2009; 249(3):488-95.

(28.) Richmond TS, Kauder D, Strumpf N, Meredith T. Characteristics and outcomes of serious traumatic injury in older adults. J Am Geriatr Soc. 2002; 50(2):215-22.

Sanjit R. Konda, M.D., Rachel Seymour, Ph.D., Arthur Manoli, III, M.D., Jordan Gales, B.S., Madhav A. Karunakar, M.D., and the Carolinas Trauma Network Research Group

Sanjit R. Konda, M.D., NYU Hospital for Joint Diseases, and Jamaica Hospital Medical Center, New York, New York. Arthur Manoli, III, M.D., and Jordan Gales, M.D., NYU Hospital for Joint Diseases New York, New York. Rachel Seymour, Ph.D., and Madhav A. Karunakar, M.D., Carolinas Medical Center, Charlotte, North Carolina.

Correspondence: Madhav A. Karunakar M.D., Carolinas Trauma Network Research Group, Carolinas Medical Center, 1025 Morehead Drive, Suite 300, Charlotte, North Carolina 28204.
Table 1 Low-Energy Geriatric and Middle-Age Trauma Patients
(Carolinas Cohort)

Development of the [STTGMA.sub.LE]: Evaluation of the Predictive
Capacity and Calibration for In-Hospital Mortality of 4 Different
Models using Variables from 4 Host Factors (Age, Pre-Existing
Conditions, ED Physiologic Signs, and Anatomic Injuries).

            Host Factors found to be independent predictors
                    of mortality for each model
                   (characteristic of variable)
                                               ED
                               Pre-existing    Physiologic
            Age                Conditions      Signs

Model 1     Age (continuous)   CCI (ordinal)   GCS
                                               (continuous)

Model 2 *   Age (continuous)   CCI (ordinal)   GCS
                                               (continuous)

Model 3     Age (continuous)   CCI (ordinal)   GCS(RTS
                                               ordinal)

Model 4     Age (continuous)   CCI (ordinal)   GCS(RTS
                                               ordinal), SBP
                                               (RTS ordinal)

              Host Factors found to be independent predictors
                        of mortality for each model
                        (characteristic of variable)

            Anatomic Injuries                        AUROC   H-L

Model 1     Major blood vessel injury                0.87    10.92
            (dichotomous), any chest injury
            (dichotomous), any intracranial injury
            (dichotomous)

Model 2 *   Head and neck injury (AIS ordinal),      0.89    8.34
            chest injury (AIS ordinal)

Model 3     Major blood vessel injury                0.86    16.20
            (dichotomous), any intracranial injury
            (dichotomous)

Model 4     Head and neck injury (AIS ordinal),      0.88    8.59
            chest injury (AIS ordinal), extremity
            injury (AIS ordinal)

CCI = Charlson Comorbidity Index; GCS = Glasgow Coma Scale; RTS =
Revised Trauma Score; SBP systolic blood pressure; AIS = Abbreviated
Injury Scale; AUROC = area under receiver operating characteristic
curve. [STTGMA.sub.LE] = Score for Trauma Triage in the Geriatric and
Middle-Aged.

Table 2 High-Energy Geriatric and Middle-Age Trauma Patients
(Carolinas Cohort)

Development of the [STTGMA.sub.HE]: Evaluation of the Predictive
Capacity and Calibration for In-Hospital Mortality of 4 Different
Models using Variables from 4 Host Factors (Age, Pre-Existing
Conditions, ED Physiologic Signs, and Anatomic Injuries).

            Host Factors found to be independent predictors of
                      mortality for each model
                    (characteristic of variable)

                                              ED
                               Pre-existing   Physiologic
            Age                Conditions     Signs

Model 1     Age (continuous)   N/A            GCS
                                              (continuous)

Model 2 *   Age (continuous)   N/A            GCS
                                              (continuous)

Model 3     Age (continuous)   N/A            GCS(RTS
                                              ordinal), (SBP

                                              RTS ordinal)

Model 4     Age (continuous)   N/A            GCS(RTS
                                              ordinal), SBP
                                              (RTS ordinal)

             Host Factors found to be independent predictors of
                         mortality for each model
                       (characteristic of variable)

            Anatomic Injuries                       AUROC   H-L

Model 1     Any intracranial injury (dichotomous)   0.93    4.99

Model 2 *   Head and neck injury (AIS ordinal),     0.96    4.23
            chest injury (AIS ordinal), extremity
            injury (AIS ordinal)

Model 3     Chest injury (dichotomous), any         0.93    5.33
            intracranial injury (dichotomous)

Model 4     Head and neck injury (AIS ordinal       0.95    5.69
            value), chest injury (AIS ordinal
            value), extremity injury (AIS ordinal
            value)

GCS = Glasgow Coma Scale; RTS = Revised Trauma Score; SBP systolic
blood pressure; AIS = Abbreviated Injury Scale; AUROC = area under
receiver operating characteristic curve, [STTGMA.sub.HE]--Score for
Trauma Triage in the Geriatric and Middle-Aged.

Table 3 Comparison of Variables Found to be Independent Predictors
of Mortality in Middle-Age and Geriatric Trauma Patients (Variables
Comprising the STTGMA)

          [STTGMA.sub.LE]    P-value   [STTGMA.sub.HE]    P-value
            OR (95% CI)                  OR (95% CI)

Age       1.05 (1.03-1.07)   < 0.01    1.08 (1.03-1.17)   < 0.01
CCI       1.28 (1.14-1.44)   < 0.01           --            --
GCS       0.72 (0.69-0.76)   < 0.01    0.69 (0.64-0.75)   < 0.01
AIS-HN    1.67 (1.49-1.87)   < 0.01    1.77 (1.44-2.21)   < 0.01
AIS-CHS   1.52 (1.19-1.92)   < 0.01    1.51 (1.20-1.90)   < 0.01
AIS-EXT          --            --      1.59 (1.14-2.21)   < 0.01

STTGMA = Score for Trauma Triage in the Geriatric and Middle-Aged;
LE-GMTP = low-energy geriatric and middle-age trauma patient;
HE-GMTP= high-energy geriatric and middle age trauma patient; CCI =
Charlson Comorbidity Index; GCS = Glasgow Coma Scale; AIS =
Abbreviated Injury Score; HN = Head & Neck; CHS = Chest; EXT =
Extremity.

Table 4 Difference in STTGMA Variables between the NTDB and Carolinas
Cohorts and Comparison of STTGMA versus TRISS in the NTDB
[STTGMA.sub.LE]

[STTGMA.sub.LE]      NTDB (N = 59,965)    Carolinas (N = 2,387)

                    Mean [+ or -] 1 SD     Mean [+ or -] 1 SD

Age                 74.7 [+ or -] 10.6     75.9 [+ or -] 11.1
GCS                 14.26 [+ or -] 2.34    14.12 [+ or -] 2.50
AIS Chest           0.48 [+ or -] 1.09     0.20 [+ or -] 0.72
AIS Head and Neck   1.58 [+ or -] 1.98     1.64 [+ or -] 1.86
CCI                 0.66 [+ or -] 1.17     0.97 [+ or -] 1.39
STTGMAle (AUROC)           0.83
TRISS (AUROC)              0.80

[STTGMA.sub.HE]      NTDB (N = 97,034)    Carolinas (N = 1,434)

                    Mean [+ or -] 1 SD     Mean [+ or -] 1 SD

Age                  67.9 [+ or -] 9.4      66.7 [+ or -] 9.2
GCS                 14.06 [+ or -] 2.91        14.09 2.85
AIS Chest           1.37 [+ or -] 1.61     1.03 [+ or -] 1.44
AIS Head and Neck    1.54 [+ or -] 2.0      0.85 [+ or -] 1.5
AIS Extremities     1.38 [+ or -] 1.20     0.93 [+ or -] 1.22
STTGMAHE (AUROC)           0.86
TRISS (AUROC)              0.85

[STTGMA.sub.LE]

                    P-value

Age                 < 0.01
GCS                  <0.01
AIS Chest           < 0.01
AIS Head and Neck    0.13
CCI                 < 0.01
STTGMAle (AUROC)    < 0.01
TRISS (AUROC)

[STTGMA.sub.HE]

                    P-value

Age                 < 0.01
GCS                  0.70
AIS Chest           < 0.01
AIS Head and Neck   < 0.01
AIS Extremities     < 0.01
STTGMAHE (AUROC)    < 0.01
TRISS (AUROC)

[STTGMA.sub.HE]--Score for Trauma Triage in the Geriatric and
Middle-Aged--High Energy; [STTGMA.sub.LE]--Score for Trauma Triage
in the Geriatric and Middle-Aged--Low Energy; TRISS--Trauma
Score-Injury Severity Score; NTDB = National Trauma Databank; GCS =
Glasgow Coma Scale; AIS = Abbreviated Injury Scale; CCI = Charlson
Comorbidity Index; SD = standard deviation.

Table 5 Risk Stratification Groups for [STTGMA.sub.LE] and
[STTGMA.sub.HE] and Estimated Reduction in Direct Hospital Costs Per
Year with Palliative Care Consultations

                              STTGMA 0              STTGMA 1
Inpatient mortality           0.0-0.9               1.0-2.9
risk (%)
                         LE        HE         LE           HE

Patient distribution     527      776     787(33.0)    206(17.2)
(%)                    (22.1)    (64.6)

Mortality (%)          1 (0.2)   1(0.1)    15(1.9)       2(1.0)

RR Mortality (95%        --        --        10.0         7.5
CI)                                       (1.3-75.8)   (0.7-82.7)

P-value                  --        --        0.03         0.10

Estimated reduction      --        --         --           --
in net hospital
costs/year with PCC
($)

Estimated reduction      --        --         --           --
in direct hospital
costs/year with PCC
($)

                               STTGMA 2               STTGMA 3
Inpatient mortality            3.0-9.9                10.0-100.0
risk (%)
                          LE        HE          LE          HE

Patient distribution     674        115        399         104
(%)                     (28.2)     (9.6)      (16.7)      (8.7)

Mortality (%)          36 (5.3)   10(8.7)   131 (32.8)   39(37.5)

RR Mortality (95%        28.2      67.5       173.0        291
CI)                     (3.9-      (8.7-      (24.3-      (40.4-
                        204.6)    522.2)     1,232.0)    2,095.8)

P-value                 <0.01      <0.01      <0.01       <0.01

Estimated reduction    483,463    86,593     402,858     110,169
in net hospital
costs/year with PCC
($)

Estimated reduction    314,684    56,790     274,369      75,413
in direct hospital
costs/year with PCC
($)

[STTGMA.sub.HE]-Score for Trauma Triage in the Geriatric and Middle-
Aged--High Energy; [STTGMA.sub.LE]-Score for Trauma Triage in the
Geriatric and Middle-Aged--Low Energy; RR = Relative Risk; CI =
Confidence Interval; PCC = Palliative Care Consultation.
COPYRIGHT 2016 J. Michael Ryan Publishing Co.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Konda, Sanjit R.; Seymour, Rachel; Manoli, Arthur, III; Gales, Jordan; Karunakar, Madhav A.
Publication:Bulletin of the NYU Hospital for Joint Diseases
Date:Oct 1, 2016
Words:5946
Previous Article:Early complications associated with the Thompson approach to the proximal radius.
Next Article:Permanent motor function loss by delayed treatment of Peroneal intraneural Ganglion.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters