Disclosure Information: Nothing to

Presented at the Western Surgical AsLake City, UT, November 2013.

Received November 11, 2013; ReDecember 30, 2013.From the Department of Surgery,Burns, and Acute Care Surgery, UTuscon, AZ.Correspondence address: Bellal JoSurgery, Division of Trauma, CriticalUniversity of Arizona, 1501 N CampTucson, AZ 85724. email: bjoseph@

ª 2014 by the American College of S

Published by Elsevier Inc.

WESTERN SURGICAL ASSOCIATION ARTICLES

Validating Trauma-Specific Frailty Index forGeriatric Trauma Patients: A ProspectiveAnalysis

Bellal Joseph, MD, FACS, Viraj Pandit, MD, Bardiya Zangbar, MD, Narong Kulvatunyou, MD, FACS,Andrew Tang, MD, FACS, Terence O’Keeffe, MBChB, FACS, Donald J Green, MD, FACS,Gary Vercruysse, MD, FACS, Mindy J Fain, MD, Randall S Friese, MD, FACS, Peter Rhee, MD, FACS

BACKGROUND: The Frailty Index has been shown to predict discharge disposition in geriatric patients. Theaim of this study was to validate the modified 15-variable Trauma-Specific Frailty Index(TSFI) to predict discharge disposition in geriatric trauma patients. We hypothesized thatTSFI can predict discharge disposition in geriatric trauma patients.

STUDY DESIGN: We performed a 2-year (2011e2013) prospective analysis of all geriatric trauma patientspresenting to our Level I trauma center. Patient discharge disposition was dichotomized intounfavorable (discharge to skilled nursing facility or death) and favorable (discharge to homeor rehabilitation center) discharge disposition. Patients were evaluated using the developed15-variable TSFI. Multivariate logistic regression was performed to identify factors thatpredict unfavorable discharge disposition.

RESULTS: A total of 200 patients were enrolled for validation of TSFI. Mean age was 77 � 12.1 years,median Injury Severity Score was 15 (interquartile range [IQR] 9 to 20), median GlasgowComa Scale score was 14 (IQR 13 to 15), and median Frailty Index score was 0.20 (IQR0.17 to 0.28); 29.5% (n ¼ 59) patients had unfavorable discharge. After adjusting for age,sex, Injury Severity Score, Head Abbreviated Injury Scale, and vitals on admission, FrailtyIndex (odds ratio ¼ 1.5; 95% CI, 1.1e2.5) was the only significant predictor for unfavorabledischarge disposition. Age (odds ratio ¼ 1.2; 95% CI, 0.9e3.1; p ¼ 0.2) was not predictiveof unfavorable discharge disposition.

CONCLUSIONS: The 15-variable TSFI is an independent predictor of unfavorable discharge disposition ingeriatric trauma patients. The Trauma-Specific Frailty Index is an effective tool that can aidclinicians in planning discharge disposition of geriatric trauma patients.

LEVEL OF EVIDENCE: II Prognostic Studies�Investigating the Effect of a Patient Characteristic on the Outcome ofDisease. (J Am Coll Surg 2014;219:10e18. � 2014 by the American College of Surgeons)

Geriatric patients have a decreased physiologic reserveand a unique physiologic vulnerability that makes clinicaldecision making in these patients challenging.1,2 As theUS population continues to age, there has been an

disclose.

sociation 121st Scientific Session, Salt

vised December 2, 2013; Accepted

Division of Trauma, Critical Care,niversity of Arizona Medical Center,

seph, MD, FACS, Department ofCare, Burns, and Acute Care Surgery,bell Ave, Rm 5411, PO Box 245063,surgery.arizona.edu

10urgeons

increase in the number of geriatric patients requiringtrauma care and admissions to our trauma centers.2,3

Predicting discharge disposition is an importantcomponent in the management of trauma patients andbegins immediately after hospital admission. Early under-standing of the discharge disposition can help in commu-nication with family and mobilizing hospital resources.Several clinical assessment tools have been used to predictadverse outcomes in trauma patients, however, none ofthese taken into account the altered physiologic reservein geriatric trauma patients.4-9 In a previous study atour institution, we demonstrated that a 50-variableFrailty Index was an independent predictor of unfavor-able discharge disposition in geriatric trauma patients.2

Frailty Index was superior to age and Injury SeverityScores for assessing adverse outcomes. However, the

ISSN 1072-7515/14/$36.00

http://dx.doi.org/10.1016/j.jamcollsurg.2014.03.020

Abbreviations and Acronyms

GCS ¼ Glasgow Coma Scaleh-AIS ¼ Head Abbreviated Injury ScaleISS ¼ Injury Severity ScoreROC ¼ receiver operating characteristicTSFI ¼ Trauma-Specific Frailty Index

Vol. 219, No. 1, July 2014 Joseph et al Geriatric Trauma-Specific Frailty Index 11

50-variable Frailty Index was an extensive and time-consuming questionnaire that was difficult to implementin the acute setting of trauma. To facilitate the clinicalimplementation of frailty in trauma, we developed amodified 15-variable Trauma-Specific Frailty Index(TSFI) using the variables from the 50-variable FrailtyIndex.The aim of this study was to validate the modified 15-

variable TSFI to predict discharge disposition in geriatrictrauma patients. We hypothesized that TSFI can predictdischarge disposition in geriatric trauma patients.

METHODSAfter obtaining approval from the University of ArizonaInstitutional Review Board, we performed a 2-year(June 2011 to July 2013) prospective observational studyof consecutive trauma patients older than 65 years pre-senting to our Level I trauma center. Only patientswith in-hospital admission were included. Patients trans-ferred from a rehabilitation center, skilled nursing facility,or other institutions, intubated or nonresponsive patientswith or without family members for whom Frailty Indexscore could not be calculated, and patients who did notconsent for enrollment were excluded.

Data points

We recorded the following data points for each patient:patient demographics (age, sex, race, and ethnicity);injury characteristics (type and mechanism of injury);vital parameters on presentation (Glasgow Coma Scale[GCS] score, systolic blood pressure, heart rate, tempera-ture); hospital and ICU lengths of stay, and dischargedisposition. The Injury Severity Score (ISS) and headAbbreviated Injury Scale (h-AIS) score were obtainedfrom the trauma registry.

Outcomes measure

We categorized patients into 2 groups based on theirdischarge disposition: favorable discharge and unfavorabledischarge. Favorable discharge was defined as discharge tohome or rehabilitation center and unfavorable defined asdischarge to a skilled nursing facility or in-hospital mor-tality. Our primary outcomes measure was unfavorabledischarge disposition.

Development of Trauma-Specific Frailty Index

We enrolled 100 consecutive geriatric trauma patients us-ing the 50-variable Frailty Index for development of theTSFI. Univariate analysis was performed to identify asso-ciations among variables in the 50-variable Frailty Indexfor development of unfavorable discharge disposition.2

Fifteen variables with the strongest association for devel-opment of unfavorable discharge disposition were selectedto develop the TSFI (Appendix; available online). Wethen enrolled 200 consecutive trauma patients aged olderthan 65 years to validate our Frailty Index.

Study protocol

Patients were approached by a single investigator on thefirst day of their hospital admission for enrollment in thestudy. After obtaining informed consent, the Frailty Indexwas calculated in each patient using the TSFI. In case ofintubated or nonresponsive patients, the information wasobtained from the patient’s close relatives. The variablescomprising the Frailty Index were explained to each pa-tient and it was clarified that the Frailty Index question-naire assessed the patient’s pre-injury health condition.

Statistical analysis

Data are reported as the mean � SD for continuousdescriptive variables, median (interquartile range [IQR])for ordinal descriptive variables, and the proportion forcategorical variables. To analyze data, we used the Stu-dent’s t-test for parametric variables and the Mann-Whitney U test for nonparametric variables. To assessthe association between each variable and an unfavorabledischarge disposition, we performed univariate analysis.Variables with a significant (p < 0.2) association perour univariate analysis were then used in a multivariatelogistic regression model. On multivariate logistic regres-sion analysis, variables were considered significant at p �0.05. Receiver operating characteristic (ROC) curve anal-ysis was used to identify the optimal TSFI cutoff point fordevelopment of unfavorable discharge disposition. For allof our statistical analyses, we used STATA Data Analysisand Statistical Software, version 11.0 (Stata Corp).

RESULTSA total of 200 consecutive geriatric trauma patients wereprospectively enrolled for validating the TSFI. Mean agewas 77 � 12.1 years, 72% (n ¼ 144) were male, medianISS score was 15 (IQR 9 to 20), median h-AIS score was 2(IQR 1 to 3), median GCS score was 14 (IQR 13 to 15),andmedian Frailty Index score was 0.20 (range 0.17 to 0.28).A favorable discharge disposition was found in 70.5%

(n ¼ 141); and the remaining 29.5% (n ¼ 59) had

12 Joseph et al Geriatric Trauma-Specific Frailty Index J Am Coll Surg

unfavorable discharge disposition. Figure 1 demonstratesthe distribution of the discharge dispositions of the studypopulation. Patients with unfavorable discharge disposi-tion had a higher median Frailty Index compared withpatients with favorable discharge. There was no differencein age, male sex, vitals on presentation, injury severity,and insurance status between the 2 groups (Table 1).Frailty Index score did not correlate with the age of thepatient (R2 ¼ 0.31; p ¼ 0.1; Figure 2).Patients with unfavorable discharge disposition had a

longer hospital length of stay and higher hospital costscompared with patients with favorable discharge disposi-tion. There was no difference in ICU length of stay andventilator days between the 2 groups (Table 2).On ROC curve analysis, TSFI cutoff point of 0.27 was

optimal, with sensitivity of 85% and specificity of 75% inpredicting unfavorable discharge disposition in geriatrictrauma patients. The area under the ROC curve was0.829 (95% CI, 0.774e0.884). Figure 3 demonstratesthe ROC analysis for unfavorable discharge disposition.On performing univariate analysis, TSFI score >0.27

was predictive of unfavorable discharge disposition(odds ratio ¼ 1.8; 95% CI, 1.4e3.1; p ¼ 0.001). Age(p ¼ 0.02), male sex (p ¼ 0.04), emergency departmentsystolic blood pressure (p ¼ 0.1), emergency departmentheart rate (p ¼ 0.2), GCS score (p ¼ 0.05), ISS (p ¼0.02), h-AIS (p ¼ 0.08), and mechanism of injury(p ¼ 0.04) were also significant factors associated withunfavorable discharge disposition. Table 3 demonstratesthe results of our univariate analysis for unfavorabledischarge disposition.After adjusting for factors associated with unfavorable

discharge disposition in a multivariate regression analysis,

Figure 1. Distribution of discharge dispositions.rehabilitation facility; SNiF, skilled nursing facility

TSFI score >0.27 remained an independent predictor ofunfavorable outcomes after trauma (odds ratio ¼ 1.5;95% CI, 1.1e2.5; p ¼ 0.001). Patient age (p ¼ 0.2),ISS (p ¼ 0.1), GCS score (p ¼ 0.2), h-AIS score (p ¼0.1), systolic blood pressure (p ¼ 0.2), and mechanismof injury (p ¼ 0.3) were not associated with unfavorabledischarge disposition. Table 4 demonstrates the results ofour multivariate logistic regression analysis for unfavor-able discharge disposition.After stratifying patients into 3 groups based on age

(65 to 74 years, 75 to 84 years, and older than 85 years),TSFI score >0.27 remained an independent predictor forunfavorable discharge disposition in each age group.Table 5 demonstrates the subanalysis for factors predict-ing unfavorable discharge disposition based on each agegroup.

DISCUSSIONOur study is the first to develop and validate aTSFI as a clin-ical assessment tool to predict unfavorable discharge dispo-sition in geriatric trauma patients. In our study, patientswith TSFI score of >0.27 were 1.5 times more likely tohave an unfavorable discharge disposition compared withpatients with low TSFI scores. Additionally, we demon-strated that, trauma-specific frailty is superior to the tradi-tionally used predictors, such as age, ISS, h-AIS score, andGCS score to predict unfavorable discharge disposition.The Trauma-Specific Frailty Index should be implementedin the emergency department as a clinical assessment toolfor geriatric trauma patients to help allocate the appro-priate resources and allow for better communication withpatients’ families.

FO, favorable discharge disposition; Rehab,; UFO, unfavorable discharge disposition.

Table 1. Demographics

Favorable outcomes (n ¼ 141) Unfavorable outcomes (n ¼ 59) p Value

Demographics

Age, y, mean � SD 76.5 � 7.5 77.1 � 7.6 0.4

65e74 y, % 51.8 49.2 0.2

75e84 y, % 34.7 35.6 0.4

Older than 85 y, % 13.5 15.2 0.5

Male, % 72.3 71.1 0.2

Whites, % 82.2 86.4 0.4

Mechanism of injury

Fall, % 57.4 59.3 0.1

MVC, % 30.5 27.1 0.3

Insured, % 94.5 95.9 0.3

FI, median (IQR) 0.18 (0.12e0.20) 0.28 (0.26e0.38) 0.01

Vital parameters

ED SBP, mean � SD 145.5 � 27.8 144.4 � 28.4 0.6

ED HR, mean � SD 85.9 � 13.9 86.9 � 19.6 0.9

ED temperature, mean � SD 36.2 � 0.8 36.6 � 0.8 0.8

GCS, median (IQR) 14 (3e15) 13 (3e15) 0.1

Injury severity parameters

ISS, median (IQR) 15 (8e18) 16 (9e21) 0.09

Head AIS, median (IQR) 2 (1e3) 2 (2e3) 0.8

AIS, Abbreviated Injury Scale; ED, emergency department; FI, Frailty Index; GCS, Glasgow Coma Scale; HR, heart rate; IQR, interquartile range; ISS, InjurySeverity Score; MVC, motor-vehicle collision; SBP, systolic blood pressure.

Vol. 219, No. 1, July 2014 Joseph et al Geriatric Trauma-Specific Frailty Index 13

In recent years, several studies in a variety of surgicalspecialties have been published implementing the conceptof frailty as an assessment tool to predict adverse

Figure 2. Distribution of

outcomes in geriatric patients.10-13 However, there iswide variability in the existing frailty assessment tools,as there is no well-defined standard tool for assessing

Frailty Index vs age.

Table 2. Length of Stay and Hospital Cost

Favorable outcomes (n ¼ 141) Unfavorable outcomes (n ¼ 59) p Value

Length of stay

Hospital LOS, d, mean � SD 3.6 � 2.2 5.8 � 3.7 0.04

ICU LOS, d, mean � SD 2.1 � 1.5 2.4 � 1.4 0.5

Ventilator days, mean � SD 1.1 � 0.3 1 � 0.7 0.2

Cost

Hospital costs, $ 18,248 � 13,018 24,479 � 15,105 0.01

Hospital charges, $ 48,235 � 30,371 59,389 � 44,023 0.02

LOS, length of stay.

14 Joseph et al Geriatric Trauma-Specific Frailty Index J Am Coll Surg

frailty. In our previous study, we used the 50-variableFrailty Index derived from the Canadian Study for Healthand Aging because we wanted to assess the utility of theestablished Frailty Index in trauma patients.2 We demon-strated that the 50-variable Frailty Index can be imple-mented in trauma patients with results similar to thoseof nontrauma and nonsurgical patients. However, the50-variable Frailty Index was time consuming and its

Figure 3. Receiver operating curve analysis for unf

clinical implementation was not feasible in the acutesetting of trauma. As time is of great importance in themanagement of trauma patients, early and quick assess-ment of frailty in the emergency department is warranted.Therefore, we developed and validated the TSFI withfewer variables, making it easy to implement, morefeasible clinically, and a significant predictor for unfavor-able discharge disposition.

avorable outcomes. AUC, area under the curve.

Table 3. Univariate Analysis for Unfavorable DischargeDisposition

Odds ratio 95% CI p Value

Frailty Index >0.27 1.8 1.4e3.1 0.001

Demographics

Age

75e84 y 1.4 1.1e2.4 0.02

Older than 85 y 1.09 1.2e1.8 0.04

Male 1.2 1.1e2.8 0.04

Whites 1.09 0.9e3.1 0.4

Hispanics 1.5 0.3e2.6 0.5

Physiological parameters

ED SBP 1.2 0.6e1.8 0.1

ED HR 1.1 0.8e2.4 0.2

ED RR 1.09 0.5e3.2 0.6

ED temperature 1.04 0.2e2.9 0.9

ED GCS score 1.3 1.1e4.2 0.05

Injury severity parameters

ISS

9�15 1.4 1.2e2.6 0.03

>16 1.2 1.1e1.8 0.02

Head AIS 1.6 0.9e2.1 0.08

Mechanism of injury 1.08 1.01e1.6 0.04

Insurance 1.3 0.8e3.6 0.4

AIS, Abbreviated Injury Scale; ED, emergency department; GCS, GlasgowComa Scale; HR, heart rate; ISS, Injury Severity Score; RR, respiratory rate;SBP, systolic blood pressure.

Table 4. Multivariate Logistic Regression for UnfavorableDischarge Disposition

Odds ratio 95% CI p Value

Frailty Index >0.27 1.5 1.1e2.5 0.001

Demographics

Age

75e84 y 1.2 0.9e3.1 0.2

Older than 85 y 1.06 0.8e1.6 0.1

Male 1.1 0.6e2.1 0.4

Physiological parameters

ED SBP 1.1 0.5e2.6 0.2

ED HR 1.04 0.9e1.5 0.5

ED GCS score 1.2 0.09e3.5 0.2

Injury severity parameters

ISS

9e15 1.2 0.8e2.1 0.1

>16 1.1 0.6e1.5 0.09

Head AIS 1.4 0.9e2.8 0.1

Mechanism of injury 1.05 0.8e1.3 0.3

AIS, Abbreviated Injury Scale; ED, emergency department; GCS, GlasgowComa Scale; HR, heart rate; SBP, systolic blood pressure.

Vol. 219, No. 1, July 2014 Joseph et al Geriatric Trauma-Specific Frailty Index 15

Several studies have demonstrated a significant associa-tion between frailty and adverse outcomes in geriatric pa-tients.13-17 In our study using the 15-variable TSFI, wefound that frail patients were more likely to have an un-favorable discharge disposition. The 15 variables thatconstitute the TSFI have been individually shown to pre-dict adverse outcomes. Robinson and colleagues demon-strated that patients with a diminished functional abilityto walk stairs were more likely to be discharged to anacute care facility.10 Makary and colleagues used theFried’s frailty score (weakness, weight loss, exhaustion,low physical activity, and walking speed) and demon-strated that frailty was an independent predictor fordischarge to a skilled nursing facility.11 In another study,Lee and colleagues defined frailty based on activity ofdaily living, ambulation, and presence of dementia, andalso found the Frailty Index to be an independent predic-tor of discharge to a rehabilitation center.14 Although allof these studies have been in nontrauma patients under-going elective or emergency general surgery operations,they support the 15 variables used to develop our TSFI.We believe that the implementation of the TSFI will pro-vide a uniform frailty scoring system based on previously

validated variables and can independently predict unfa-vorable discharge disposition.In our previous study using the 50-variable modified

Rockwood Frailty Index, we found that frail trauma pa-tients were 1.3 times more likely to have an unfavorabledischarge disposition compared with nonfrail patients.2

The result of our current study using the 15-variableTSFI score are comparable with results of our previousstudy demonstrating similar predictive value betweenthe 50-variable Frailty Index and the 15-variable TSFI.In our study, patients with unfavorable discharge

disposition had a higher median TSFI score (0.18; range0.12 to 0.20) compared with patients with favorabledischarge disposition (0.28; range 0.26 to 0.38). Theoptimal cutoff point for development of unfavorabledischarge disposition was found to be 0.27(sensitivity ¼ 85% and specificity ¼ 75%). The resultsof our study are consistent with other studies publishedin the literature that have defined a Frailty Index cutoffof 0.25 in nontrauma and nonsurgical patients.15,16 Geri-atric trauma patients with a TSFI cutoff score of >0.27are more likely to have an unfavorable discharge disposi-tion; early assessment of frailty using the TSFI can aid incommunication with family and hospital resource mobi-lization for discharge disposition in this cohort ofpatients.Studies assessing outcomes in trauma patients have al-

ways focused on age (older than 65 years) and injuryseverity as predictors for adverse outcomes.17 However,none of these studies have taken into account the altered

Table 5. Multivariate Logistic Regression for Unfavorable Discharge Disposition Categorized by Age

Age 65e74 years Age 75e84 years Age older than 85 years

OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) p Value

Frailty Index >0.27 1.6 (1.2e2.4) 0.01 1.2 (1.1e1.6) 0.01 1.1 (1.04e1.9) 0.02

Male 1.1 (0.8e1.5) 0.2 1.8 (0.6e4.3) 0.1 1.3 (0.6e3.5) 0.4

ED SBP 1.2 (0.6e1.9) 0.1 1.5 (0.4e2.3) 0.5 1.4 (0.9e2.1) 0.2

ED HR 1.6 (0.4e3.1) 0.8 1.3 (0.7e2.6) 0.4 1.1 (0.5e3.9) 0.8

ED GCS score 1.1 (0.2e3.4) 0.5 1.4 (0.8e1.5) 0.1 1.2 (0.7e4.9) 0.6

ISS

9e15 1.1 (0.9e1.9) 0.2 1.5 (0.8e2.8) 0.1 1.4 (0.8e2.5) 0.1

>16 1.4 (0.7e2.6) 0.1 1.2 (0.5e1.5) 0.3 1.8 (0.9e1.9) 0.09

Head AIS 1.2 (0.9e1.8) 0.1 1.3 (0.7e1.5) 0.2 1.2 (0.9e2.1) 0.09

AIS, Abbreviated Injury Scale; ED, emergency department; HR, heart rate; ISS, Injury Severity Scale; OR, odds ratio; SBP, systolic blood pressure.

16 Joseph et al Geriatric Trauma-Specific Frailty Index J Am Coll Surg

physiologic reserve among geriatric trauma patients. Inour previous study, we found the Frailty Index was supe-rior to age and injury severity in predicting unfavorabledischarge disposition. The results of our current studyvalidate the findings of our previous studies and demon-strate the TSFI to be the only independent predictor forunfavorable discharge. Similarly, Shah and colleaguesfound no difference in functional outcomes based onage in patients undergoing rehabilitation.9 We believethat chronologic age of the patient and injury severityshould not be used as a determinant of outcomes in geri-atric patients, but rather a detailed assessment of the pa-tient’s physiologic age and existence of frailty syndromeis required when predicting adverse outcomes in geriatricpatients. The TSFI should be made an integral part of thetrauma assessment of geriatric trauma patients in theemergency department.Chronologic age is commonly considered as a proxy to

define the physiologic vulnerability among individuals,18-20

however, there is wide variability in the age at which in-dividuals start deteriorating physiologically. Some indi-vidual appear to be frail at age 70 years, and othersreach this state at 90 years of age. The concept of frailtywas developed to help better understand and define thisheterogeneity in functional decline among aging pa-tients.18-23 In our study, we found no correlation betweenfrailty and chronologic age of the patient. Althoughstudies have shown association between frailty andage,18-20 in recent years several studies have found frailtyto be independent of the chronologic age of an individ-ual.21-23 Studies have propagated several mechanisms toexplain the differences in outcomes between frail andnonfrail patients, but the exact mechanism remains un-clear.19 In trauma patients, worse outcomes have beenshown to be associated with differences in sex, race,ethnicity, hormonal variability, and even social sta-tus.24-26 We believe that additional studies are requiredto assess the impact of sex and hormonal variability at

the molecular and biochemical level to better understandthe relation between frailty and chronological age amongtrauma patients.Our current study had several limitations. We did not

evaluate the impact of frailty on long-term functionaloutcomes and quality of life. Second, our results wereobtained at a single academic medical center and there-fore might not be generalizable. Third, we have a smallsample size for validating our TSFI.

CONCLUSIONSThe 15-variable TSFI is an independent predictor of unfa-vorable discharge disposition in geriatric trauma patients.The TSFI is an effective tool that can aid clinicians in plan-ning discharge disposition of geriatric trauma patients.Clinical implementation of the TSFI is warranted.

Author Contributions

Study conception and design: Joseph, Pandit, Zangbar,Fain, Friese, Rhee

Acquisition of data: Joseph, Pandit, Zangbar, Kulvatunyou,Tang

Analysis and interpretation of data: Pandit, Zangbar,Kulvatunyou, Fain, Friese, Rhee

Drafting of manuscript: Joseph, Pandit, Zangbar, Tang,O’Keeffe, Green, Vercruysse

Critical revision: Joseph, Tang, O’Keeffe, Green,Vercruysse, Fain, Friese, Rhee

REFERENCES

1. Polanczyk CA, Marcantonio E, Goldman L, et al. Impact ofage on perioperative complications and length of stay in pa-tients undergoing non-cardiac surgery. Ann Intern Med2001;134:637e643.

2. Joseph B, Pandit V, Rhee P, et al. Predicting hospital dischargedisposition in geriatric trauma patients: is frailty the answer?J Trauma Acute Care Surgery 2014;76:196e200.

3. USDepartment of Health and Human Services, Administrationon Aging. A Profile of Older Americans: 2010. Available at:

Vol. 219, No. 1, July 2014 Joseph et al Discussion 17

http://www.aoa.gov/aoaroot/aging_statistics/profile/2010/docs/2010profile.pdf. Accessed November 8, 2013.

4. Boyd CM, Darer J, Boult C, et al. Clinical practice guidelinesand quality of care for older patients with multiple comorbiddiseases: implications for pay for performance. JAMA 2005;294:716e724.

5. Woods NF, LaCroix AZ, Gray SL, et al. Frailty: emergenceand consequences in women aged 65 and older in the women’shealth initiative observational study. J Am Geriatr Soc 2005;53:1321e1330.

6. Fried LP, Kronmal RA, Newman AB, et al. Risk factors for 5-year mortality in older adults: the cardiovascular health study.JAMA 1998;279:585e592.

7. Zafonte R, Flora H, Nancy M, et al. Relationship betweenGlasgow Coma Scale and functional outcome. Am J PhysMed Rehabil 1996;75:364e369.

8. Foreman BP, Caesar RR, Parks J, et al. Usefulness of theabbreviated injury score and the injury severity score in com-parison to the Glasgow Coma Scale in predicting outcome af-ter traumatic brain injury. J Trauma 2007;62:946e950.

9. ShahMK, Al-Adawi S, Burke D. Age as a predictor of functionaloutcome in anoxic brain injury. J Appl Res 2004;4:380e384.

10. Robinson T, Wallace JI, Wu DS, et al. Accumulated frailtycharacteristics predict postoperative discharge institutionaliza-tion in the geriatric patient. J Am Coll Surg 2011;213:37e42.

11. Makary MA, Segev DL, Pronovost PJ, et al. frailty as a predic-tor of surgical outcomes in older patients. J Am Coll Surg2010;210:901e908.

12. Dana HL, Karen JB, Billie JM, et al. Frail patients are atincreased risk for mortality and prolonged institutional care af-ter cardiac surgery. Circulation 2010;121:973e978.

13. Jacqueline MGW, Priyanka G, Jack WG, et al. Frailty anddelayed graft function in kidney transplant recipients. ArchSurg 2012;147:190e193.

14. Lee DH, Buth KJ, Martin BJ, et al. Frail patients are atincreased risk for mortality and prolonged institutional care af-ter cardiac surgery. Circulation 2010;121:973e978.

15. Rockwood K, Andrew M, Mitnitski A. A comparison of twoapproaches to measuring frailty in elderly people. J GerontolA Biol Sci Med Sci 2007;62:738e743.

16. Wou F, Gladman JR, Bradshaw L, et al. The predictive prop-erties of frailty-rating scales in the acute medical unit. AgeAgeing 2013;42:776e781.

17. Hashmi A, Ibrahim-Zada I, Rhee P, et al. Predictors of mor-tality in geriatric trauma patients: a systematic review andmeta-analysis. J Trauma Acute Care Surg 2014;76:894e901.

18. Mitnitski AB, Graham JE, Mogilner AJ, Rockwood K. Frailty,fitness and late-life mortality in relation to chronological andbiological age. BMC Geriatr 2000;2:1.

19. Bergman H, Ferrucci L, Guralnik J, et al. Frailty: an emergingresearch and clinical paradigmdissues and controversies.J Gerontol A Biol Sci Med Sci 2007;62:731e737.

20. Fried LP, Tangen CM, Walston J, et al. Frailty in older adultsevidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146eM157.

21. Walston J, McBurnie MA, Newman A, et al. Frailty andactivation of the inflammation and coagulation systemswith and without clinical comorbidities: results from theCardiovascular Health Study. Arch Intern Med 2002;162:2333e2341.

22. Leng SX, Cappola AR, Andersen RE, et al. Serum levels ofinsulin-like growth factor-I (IGF-I) and dehydroepiandrosterone

sulfate (DHEA-S), and their relationships with seruminterleukin-6, in the geriatric syndrome of frailty. Aging ClinExp Res 2004;16:153e157.

23. Buchner DM, Wagner EH. Preventing frail health. Clin Ger-iatr Med 1992;8:1e17.

24. Shafi S, de la Plata CM, Diaz-Arrastia R, et al. Racial dispar-ities in long-term functional outcome after traumatic braininjury. J Trauma Acute Care Surg 2007;63:1263e1270.

25. Wright DW, Kellermann AL, Hertzberg VS, et al. ProTECT:a randomized clinical trial of progesterone for acute traumaticbrain injury. Ann Emerg Med 2007;49:391e402.

26. Haider AH, Chang DC, Efron DT. Race and insurance statusas risk factors for trauma mortality. Arch Surg 2008;143:945.

Discussion

INVITED DISCUSSANT: DR CHRISTOPHER BANDY (Vaca-

ville, CA): I would like to applaud the authors of this paper forfocusing on an extremely important segment of our population.The over-85-year-olds represent the fastest growing segment of

our society and many are not as frail as people their age were 20years ago. Many are very active, therefore placing themselves atgreater risk sometimes from mechanisms of injury that were un-

heard of in previous decades.Reviewing the 15 variables you have selected for the Trauma Spe-

cific Frailty Index (TSFI), one might consider at least 5, and up to10, of these variables to be somewhat subjective. Additionally, as

you have stated in your methods, if a patient is obtunded (ordemented), an investigator reviews these variables with the patient’snext of kin, possibly introducing further subjectivity (or maybe even

decreasing it). For example, many spouses disagree with one anotheron many topics. I could imagine some men answering that they werenot “sexually active,” yet their spouses answering to the contrary. It

depends on perspective; and how many years they’ve been married.How many patients in your population were obtunded or other-

wise unable to participate in this examination? What are your

thoughts on the subjectivity of some of these variables and whatsteps, if any, did you take to minimize this potential bias? Sincecompletion of this paper, have any of you considered applyingthe same univariate analysis used to define the 15 variables of

TSFI to further refine, and possibly strengthen, the association be-tween frailty and discharge disposition?

Last, in a prospective study by a urology group from Emory in

last month’s JACS, the investigators used the “Hopkins FrailtyScore” (in addition to an exhaustive list of subjective and objectivescoring systems and biochemical markers) to question the risk of

postoperative complications in 189 patients. The percentage of“intermittently frail” and “frail” patients in their population was26.5%, very similar to the 29.5% of patients in your study with “un-favorable discharge” potential. Interestingly, they included patients

over the age of 18 (your population is defined by age 65 andover), postulating that frailty is “not a function of chronological age.”

I am certain that you have cared for many “frail” patients under

the age of 65. Have you thought about expanding your inclusioncriteria? Because age was not a determining factor in “unfavorable

Appendix. 15-Variable Trauma-Specific Frailty Index

Comorbidities

Cancer history

Yes 1

No 0

Coronary heart disease

Myocardial infarction 1

Coronary artery bypass grafting 0.75

Percutaneous coronary intervention 0.5

Medication 0.25

No medication 0

Dementia

Severe 1

Moderate 0.5

Mild 0.25

None 0

Daily activities

Help with grooming

Yes 1

No 0

Help managing money

Yes 1

No 0

Help doing household work

Yes 1

No 0

Help toileting

Yes 1

No 0

Help walking

Wheelchair 1

Walker 0.75

Cane 0.25

None 0

Health attitude

Feel less useful

Most time 1

Sometimes 0.5

Never 0

Feel sad

Most time 1

Sometimes 0.5

Never 0

Feel effort to do everything

Most time 1

Sometimes 0.5

Never 0

Falls

Most time 1

(Continued)

Appendix. (Continued)

Sometimes 0.5

Never 0

Feel lonely

Most time 1

Sometimes 0.5

Never 0

Function, sexually active

Yes 1

No 0

Nutrition, albumin

<3 1

>3 0

Vol. 219, No. 1, July 2014 Joseph et al Geriatric Trauma-Specific Frailty Index 17.e1