Risk Factors for Delirium in Trauma Patients:The Impact ofEthanol Use and Lack of Insurance

BERNARDINO C BRANCO, M.D.,* KENIIINABA, M.D.,* MARKO BUKUR, M.D.,t PEEP TALVING, M.D., PH.D.,*MATTHEW OLIVER, M.D.,* lEAN-STEPHANE DAVID, M.D.,t LYDIA LAM, M.D.,*

DEMETRIOS DEMETRIADES, M.D., PH.D.*

From the *Division of Trauma and Surgical Critical Care, University of Southern California, Los Angeles,California; f Division of Trauma and Critical Care, Cedars-Sinai Medical Center, Los Angeles,

California; and the fDepartment of Anesthesiology & Critical Care, Lyon-Sud Hospital,Hospital Civilis de Lyon and Claude Bernard University, Lyon, France

The purpose of this study was to examine independent risk factors, and in particular the impact ofalcohol on the development of delirium, in a cohort of trauma patients screened for ethanol in-gestion on admission to hospital. The National Trauma Databank (v. 7.0) was used to identify allpatients 18 years or older screened for ethanol on admission. Patients who developed deliriumwere compared with those who did not. Stepwise logistic regression analysis was used to identifyindependent risk factors for the development of delirium. A total of 504,839 patients withadmission ethanol levels were identified. Of those, 2,909 (0.6%) developed delirium. Patientsdeveloping delirium were significantly older, more frequently male, and more likely to sustainthermal injuries and falls. Patients developing delirium had more comorbidities includingchronic ethanol use (19.1% vs 4.5%, P < 0.001) and cardiovascular disease (21.5% vs 12.2%, P <0.001). On admission, patients developing delirium were more likely to be intoxicated with eth-anol (55.4% vs 26.5%, P < 0.001) and were more likely to be uninsured (17.8% vs 0.9%, P < 0.001). Astepwise logistic regression model identified lack of insurance, positive ethanol on admission,chronic ethanol use. Intensive Care Unit admission, age > 55 years, bums. Medicare insurance,falls, and history of cardiovascular disease as independent risk factors for the development ofdelirium. The incidence of delirium in this trauma patient cohort was 0.6 per cent. The above riskfactors were independently associated with the development of delirium. This data may behelpful in designing interventions to prevent delirium.

T HE DEVELOPMENT OF DELIRIUM during hospital ad- of delirium in trauma pafients.'̂ - ^ These studies weremission is associated with significant morbidity, limited by a small sample size and the lack of logisdc

Previous reports have documented increased compli- regression analysis to identify predictors for the de-cation rates, prolonged Intensive Care Unit (ICU) velopment of delirium.length of stay (LOS) and hospital LOS when delirium The idenfificafion of pafients at risk of developingdevelops, resulfing in an increase in hospitalizafion delirium is important as it may facilitate the inifiationand treatment costs.'^ Moreover, delirium may be of prophylacfic treatment, allow early diagnosis, anddifficult to disfinguish from sepsis, shock, or pro- provide effecfive interventions for those who developgression of traumafic brain injury, complicating the delirium despite prophylaxis with the practical goal ofdiagnosis and treatment of these conditions.^ reducing complicafions such as self-extubafion, falls,

Ethanol use is prevalent in pafients admitted to a and life-threatening agitation. The purpose of thetrauma center after injury. It is estimated that a quarter present study was to examine the prevalence of de-of pafients admitted to urban hospitals are posifive for lirium in an acutely injured pafient cohort and toethanol.^ These patients may be at significant risk of • idenfify independent risk factors for its development,developing delirium. To date, very few studies have ex-amined the risk factors associated with the development

Methods

— ~ , . , . T̂ •• , u The Nafional Trauma Databank (NTDB) of theAddress eorrespondence and repnnt requests to Kenji Inaba, . . „ , , ^_ . - , , > j ^ yy..'

M.D., Division of Trauma and Surgieal Critieal Care, University of American College of Surgeons version 7.0 was queriedSouthern California, 1200 North State Street, Room CL5100, Los to idenfify all trauma pafients 18 years or older. ThisAngeles, CA 90033-4525. E-mail: kinaba@surgery.use.edu. version of the NTDB contains deidenfified data from

621

622 THE AMERICAN SURGEON May 2011 Vol, 77

the years of 2002 to 2006, Patients with a hospitalLOS > 24 hours were evaluated and those with missingserum ethanol levels on admission were excluded toassess the role of ethanol as a risk factor for thedevelopment of delirium. Patient variables extractedincluded age, gender, ethnicity, injury mechanism,comorbidities (cardiovascular disease, diabetes melli-tus, dementia, psychiatric disease, and chronic ethanolor illicit drug use), admission vitals, Glasgow ComaScale (GCS) score and toxicology results for ethanoland illicit drugs (-1- or -) on admission, description ofinjuries (traumatic brain injury, intrathoracic and/orintra-abdominal injury, pelvic fracture, and extremityfracture). Injury Severity Seore (ISS), type of insurance,and outcomes (̂ ventilation days, ICU and hospital LOS,and mortality).

The population was divided into two groups: thosewho developed delirium and those who did not. TheInternational Classification of Diseases, 9th Revision,Clinical Modification codes (ICD-9) 290,11, 290.41,291.0, 292.81, and 293.0-1 were used to identify pa-tients who developed delirium. In the present evalua-tion, ICD-9 codes for delirium were used exclusivelyas a diagnostic tool.

Continuous variables were dichotomized using thefollowing clinically relevant cut-points: age (^ 55 vs <55 years), systolic blood pressure (SBP) on admission(< 90 vs > 90 mm Hg), GCS on admission (< S vs >8), and ISS (^ 16 vs < 16), These two groups werecompared for differences in demographics, comor-bidities, clinical data ,and outcomes using bivariateanalysis, x^ test was used to compare proportions, andunpaired Student's t test was used to compare means.For the analysis of outcomes. Analysis of Covarianceand binary logistic regression were used to adjust fordifferences between the two groups.

To identify independent risk factors for the de-velopment of delirium, factors that on bivariate anal-ysis were significant at P < 0,2, were entered in astepwise logistic regression model analysis. Positiveand negative predictive values were calculated to de-termine how well this model predicted the develop-ment of delirium. The summary data is presented as araw percentage or mean ± standard deviation. The Pvalues were significantly different at P < 0,05, Datawere analyzed using SPSS for Windows, version 16.0(SPSS, Chicago, IL).

Results

During the 5-year study period, a total of 1,503,074adult trauma patients were identified. After exclusionof 538,242 with hospital LOS < 24 hours and 459,993with missing serum ethanol levels on admission,504,839 patients (49.6%) remained for analysis andconstituted the study population. Of those, 2,909(0.6%) developed delirium and 501,930 (99.4%) didnot (Fig. 1),

Patients developing delirium were significantlyolder (53,9 ± 18.0 vs 44.8 ± 19.7, P < 0.001), morefrequently male (72.8% vs 67.4%, P < 0,001), morelikely to sustain thermal injuries (2.3% V5 1.1%, P <0,001) and, amongst blunt injuries, falls were mostcommon (41.5% vs 25.4%, P < 0.001). Patients de-veloping delirium had more comorbidities includingchronic ethanol use (19,1% vs 4,5%, P < 0,001) andcardiovascular diseases (21,5% vs 12,2%, P < 0,001). Onadmission, patients developing delirium were morelikely to be intoxicated with ethanol (55.4% vs 26.5%,P < 0,001) and illicit drugs (18,6% vs 14,3%, P < 0,001).Patients developing delirium were more severely injuredwith higher ISS (12.2 ± 10.2 vs 11.6 ± 10.7, P = 0.005)

NTDB2002 - 20061,861,779

Age £ 18 years1,503,074(80.7%)

Exclusioti

EiG. 1. Study outline.

n=998,235_ J © HLOS<24 h, n=538,242

0 No ethanol levels, n=459,993

Delirium2,909(0.6%)

Study Popuiation504,839(49.6%)

)

No Delirium501,930(99.4%)

No. 5 RISK FACTORS FOR DELIRIUM IN TRAUMA PATIENTS Branoo et al. 623

and were more likely to have a SBP < 90 mm Hg onadmission (9.1% vs 5.9%, P < 0.001). Patients de-veloping delirium were also more likely to be un-insured (17.8% vs 0.9%, P < 0.001) and to haveMedicare (24.8% vs 13.9%, P < 0.001). Comparison ofdemographics, comorbidities, and clinical data be-tween groups is presented in Table 1.

As shown in Fig. 2, the incidence of delirium variedwidely according to ethanol status on admission aswell as chronic history of ethanol use; from 0.2 percent for patients negative for ethanol to 2.8 per cent forthose with chronic history of ethanol use admitted withposidve ethanol levels.

After adjusfing for differences in age, gender, injurymechanism, comorbidities, intubation requirements,SBP, GCS, and toxicology status on admission, ISS,

associated injuries, and type of insurance betweengroups, padents who developed delirium had signifi-canfiy longer vendladon days (2.5 ± 35.1 vs 0.7 ± 10.4,P < 0.001), ICU LOS (4.0 ± 8.3 vs 2.2 ± 8.3, P <0.001), and hospital LOS (13.0 ± 12.4 vs 8.3 ± 12.6,P < 0.001). There was no significant difference inmortahty (3.4% vs 2.8%, P = 0.735) between the twogroups (Table 2). A stepwise logisfic regression modelidentified lack of insurance, the presence of ethanol onadmission, chronic ethanol use, ICU admission, age s55 years, burns. Medicare insurance, falls, and historyof cardiovascular disease as independent risk factorsfor the development of delirium (Table 3).

To determine how well our model predicted de-lirium, we calculated the positive predictive values andnegative predictive values when no risk factors for

TABLE 1. Demographics, Comorbidities, and Clinical Data Between Patient Croups

DemographicsAge (y), mean ± SDAge > 55MaleCaucasianPenetratingBurnBlunt

MVAFall

ComorbiditiesChronic ethanol useChronic drug useCardiovascular diseaseDiabetes mellitusDementiaPsychiatric diseasei~^linipnl Hâta

Positive for ethanol on admissionPositive for drug on admissionIntubated on admissionSBP on admission, mean ± SDSBP on admission < 90 mmHgGCS on admission < 8ISS, mean ± SDISS > 16Traumatic brain injuryIntrathoraeic and/or

intra-abdominal injuryPelvic fractureExtremity fractureUninsuredMedicareMedicaidOthert

Total(n = 504,839)

44.9 ± 19.728.9% (145,909)67.4% (340,402)64.4% (325,154)10.8% (54,738)

1.1% (5,634)87.0% (439,376)45.2% (228,336)25.5% (128,917)

4.6% (23,222)2.5% (12,832)

12.3% (61,864)3.9% (19,925)0.9% (4,602)3.9% (19,809)

26.6% (134,479)14.4% (72,535)5.3% (26,835)133.3 + 35.2

5.9% (28,525)23.2% (104,915)

11.6+ 10.829.9% (150,055)14.9% (75,299)14.1% (71,313)

4.8% (24,411)19.2% (96,833)

1.0% (5,128)14.0% (70,702)7.2% (36,442)

73.5% (370,867)

Delirium(n = 2909)

53.9 ± 18.043.2% (1,257)72.8% (2,119)76.5% (2,224)

8.1% (235)2.3% (66)

88.2% (2,567)33.5% (974)41.5% (1,207)

19.1% (555)5.6% (163)

21.5% (624)4.9% (143)1.2% (36)5.4% (156)

55.4% (1,613).18.6% (541)

9.3% (270)131.1 +41.39.1% (226)

22.5% (567)12.2 + 10.2

32.5% (941)11.9% (345)12.4% (362)

3.9% (113)15.1% (438)17.8% (518)24.8% (721)

8.9% (258)46.6% (1,355)

No Delirium(n = 501,930)

44.8 ± 19.728.8% (144,652)67.4% (338,283)64.3% (322,930)10.9% (54,503)

1.1% (5,568)87.0% (436,809)45.3% (227,362)25.4% (127,710)

4.5% (22,667)2.5% (12,669)

12.2% (61,240)3.9% (19,782)0.9% (4,566)3.9% (19,653)

26.5% (132,866)14.3% (71,994)5.3% (26,565)133.3 + 35.1

5.9% (28,299)23.2% (104,348)

11.6+ 10.729.9% (149,114)14.9% (74,954)14.1% (70,951)

4.8% (24,298)19.2% (96,395)0.9% (4,610)

13.9% (69,981)7.2% (36,184)

73.6% (369,512)

P

<O.OOI*<0.001*<0.001*<0.001*<O.OOI*<O.OOI*<0.051<0.001*<0.001*

<0.001*<0.001*<0.001*

0.007*0.064*

<0.001*

<0.001*<0.001*<0.001*

0.002*<0.001*

0.3940.005*0.003*

<O.OOI*0.009*

0.016*<0.001*<0.001*<0.001*

0.001*<0.001*

The P values for categorical variables were derived from x test; P values for continuous variables were derived from unpairedStudent's / test.

* P values are significantly different (P < 0.05).t Other includes self-pay, commercial plan, automobile. Blue Cross/Shield, worker's compensation, managed care organiza-

tion, and government/military insurances.SD, standard deviation; MVA, motor-vehicle accident.

624 THE AMERICAN SURGEON May 2011 Vol. 77

2.8%

- Ethanol onadmission

+ Ethanol onadmission

No chronic ethanol Chronic ethanol use + Ethanol onadmission and chronic

use

FIG. 2. Incidence of delirium according to admission ethanol status and chronic ethanol use.

TABLE 2. Outcomes Between Patient Groups

Total(n = 504,839)

Delirium(n = 2909)

No Delirium(n = 501,930)

Adjusted MeanDifference(95% CI) Adjusted P

Venttlation days, mean ± SDICU LOS, mean ± SDHospital LOS, mean ± SD

ICU admission (%)Ventilated (%)Mortality (%)

4320

2

0.72.28.4

.3%

.9%

.8%

± 10.7±8.3± 12.6

(190,722)(80,887)(14,106)

2.5 ±35.14.0 ± 8.3

13.0 ± 12.4

58.3% (1,454)30.4% (651)3.4% (100)

0.72.28.3

43.2%20.9%

2.8%

± 10.4±8.3± 12.6

(189,268)(80,236)(14,006)

1.8(1.3,2.3)1.8(1.4,2.0)4.6 (4.2, 5.2)

Adjusted OR(95% CI)

1.8 (1.6, 1.9)1.7(1.5, 1.8)1.1 (0.8, 1.3)

<0.001*<0.001*<0.001*

Adjusted P

<0.001*<0.001*

0.735

The P values were derived from Analysis of Covariance for ventilation days, ICU LOS, and hospital LOS; and from binarylogistic regression for mortality and ICU and ventilation requirements.

The P values were obtained after adjustment for age, gender, mechanism of injury, comorbidities, intubation requirements, SBPand GCS on admission, ethanol and drug levels on admission, ISS, associated injuries, and type of insurance.

* P values are significantly different {P < 0.05).CI, conftdence interval; SD, standard deviation; OR, odds ratio.

TABLE 3. Risk Factors for Development of Delirium

Step123456789

VariableUninsured+ Ethanol on admissionChronic ethanol useICU admissionAge > 55 yearsBurnsMedicareFallsCardiovascular disease

Delirium

10.1%1.2%2.4%0.8%0.9%1.2%1.0%0.9%1.0%

R'0.070.040.030.030.020.010.010.010.01

Adjusted OR(95% Cl)

52.6 (38.5, 71.4)4.9(4.1,5.1)2.3(2.1,2.6)2.1 (1.9,2.3)1.9(1.7,2.1)1.8(1.3,2.5)2.7(1.1, 1.9)1.7(1.5, 1.9)1.3(1.2,2.6)

Adjusted P

<0.001*<0.00] *<0.00]*<0.001*<0.001*<0.001*<0.001*<0.001*<0.001*

Variables entered in the regression: age s 55 years, gender, mechanism of injury, comorbidities, intubation requirements, SBP <90 mm Hg and GCS < 8 on admission, ethanol and drug levels on admission, ISS > 16, traumatic brain injury, intrathoracic and/orintra-abdominal injury, pelvic fracture, extremity fracture, type of insurance, and need for ICU admission.

A total of 389,971 (77.2%) subjects with complete data were included in the model.* P values are significantly different {P < 0.05).OR, odds ratio; Cl, confidence interval.

delirium were present, when > 2, > 3, ^ 4, and > 5risk factors were present. When no risk faetors werepresent, < 0.001 per eent of these patients developeddelirium. When s 5 factors were present, 71.4 per centof patients developed delirium (Table 4).

Discussion

The development of delirium is associated withsignificant morbidity. Previous studies have docu-mented increased complications and a prolonged LOS

No. 5 RISK FACTORS FOR DELIRIUM IN TRAUMA PATIENTS Branco et at. 625

TABLE 4. Model Prediction for Delirium Development According to the Presence and Number of Risk Factors

Risk Factors Present Events Delirium PPV (95% CI) NPV (95% CI)

No (none)> 2 factorss 3 factorsS: 4 factors> 5 factors

42/91,835365/1,75055/17168/1605/7

<0.001%20.9%32.2%42.5%71.4%

0.00 (0.00, 0.00)0.29 (0.25, 0.42)0.32 (0.26, 0.40)0.43 (0.39, 0.46)0.72 (0.35, 0.92)

0.99 (0.99, 0.99)0.99 (0.98, 0.99)0.98 (0.97, 0.99)0.99 (0.98, 0.99)0.99 (0.99, 0.99)

Risk factors: lack of insurance, positive ethanol on admission, chronic ethanol use, ICU admission, age > 55 years, burns.Medicare insurance, falls, and cardiovascular disease.

PPV, positive predictive values; NPV, negative predictive values.

when delirium develops, resuldng in increased hospitalcosts. In a study published in the late 1990s, Spies andcolleagues' evaluated a cohort of 213 cridcally illmedical padents, 39 of whom developed delirium.These patients had an increased incidence of pneu-monia, sepsis, and a prolonged ICU stay by approxi-mately 8 days when compared with those who did notdevelop delirium.' Similarly, Marik and Mohedin^prospecdvely collected 200 ethanol-dependent pa-dents, of which 20 per cent developed delirium. Thesepadents had their ICU stay increased by 4 days andtotal hospital charges increased by $10,000.^ Moreover,delirium may also be confused with a wide spectrum ofdiseases, in particular sepsis, complicadng the diagnosisand treatment of these conditions.^

In the present study, we used the NTDB, the largestnational trauma database from over 900 United Statestrauma centers. The aim was to idendfy padents whodeveloped delirium and to determine the independentrisk factors for its development. We included adulttrauma patients for whom ethanol levels on admissionwere available (positive or negative) as a history ofethanol use has been shown to significantly impact thedevelopment of delirium.^- ''• ^ We identified 2,909padents (0.6% of the inidal study populadon) whodeveloped delirium. After stepwise logistic regressionanalysis, nine independent risk factors were idendfied:1) lack of insurance, 2) the presence of ethanol onadmission, 3) chronic ethanol use, 4) ICU admission,5) age > 55 years, 6) bums, 7) Medicare insurance, 8)falls, and 9) previous history of cardiovascular disease.When no risk factors were present, only 42 patientsout of 91,835 (< 0.001%) developed delirium. Thenegative predictive value was close to 100 per cent. Onthe other hand, if five or more of the above risk factorswere present, 72 per cent of patients developed de-lirium. Although these results will require further pro-specdve validadon, they may aid in the idendficadon ofthe at-risk padent cohort allowing more aggressivemonitoring and rapid treatment once idendfied.

Lack of insurance was the strongest predictor for thedevelopment of delirium in our model. It increased therisk of developing delirium by over 50-fold. Thisfinding may be an indicator of socioeconomic status.

Uninsured padents may be more susceptible to chronicand/or acute ethanol use.̂ - '° Two recent publicationsby Rosen et al." and Salim et al.'^ also using theNTDB demonstrated worse overall outcomes amonguninsured padents after trauma. The authors hypothe-sized three possible reasons for such difference: 1)Lack of insurance may be associated with delays intreatment, 2) Uninsured patients may receive a differ-ent standard of care than insured patients, and 3) Un-insured padents may possess a lower rate of healthliteracy stemming from or resuldng in less effectivecommunieadon with their trauma care providers." Al-though the present study does not allow for an analysisof the causative factors underlying this strong relation-ship between the lack of insurance and development ofdelirium, this represents an area for possible interven-tion and further research into this association.

This study also provides quantitative support to theassociation between ethanol and delirium. The pres-ence of ethanol on admission increased the risk ofdeveloping delirium by 4-fold. If chronic ethanol usewas present, this risk increased by 12-fold. Althoughdelirium was first described as early as 1787, its as-sociation to ethanol was not established until the mid1900s.'3 Ethanol use is highly prevalent amongst in-jured patients being evaluated at a trauma center. It isesdmated that a quarter of padents admitted to urbanhospitals are posidve for ethanol.^ To date, few studieshave evaluated this association after trauma. In 2002,Lukan and colleagues'' reviewed all trauma patientsadmitted to the University of Louisville Hospitaltesdng positive for ethanol; 6 per cent of them de-veloped delirium. When patients who developed de-lirium were compared with those who did not, patientswho developed delirium were significantly older(age > 40), more likely to be Caucasian and to havesustained thermal injuries. These patients had a 3-foldhigher risk of developing delirium. Two years later, thesame group applied multivariable analysis to a cohortof 265 trauma patients. The presence of a positiveadmission blood alcohol level, elevated mean corpus-cular volume, and age greater than 45 years wereidentified as predictors of delirium. Positive admissionblood alcohol, in particular, was associated with a

626 THE AMERICAN SURGEON May 2011 Vol. 77

6-fold adjusted odds rafio of developing delirium. Theauthors concluded that alcohol disorders may haveplayed an important role in the development of de-lirium in this populafion.^

In addition to ethanol, several other factors have alsobeen shown to infiuence the development of delirium.Injury severity, blood loss, and other acute events asso-ciated with autonomie hyperactivity are likely to lowerthe threshold for the development of delirium. Age, nu-tritional status, and preexisting medical comorbidities arealso likely to infiuence the development of delirium aswell. Einally, medical management, including electrolyteimbalances that may occur during inifial resuscitafion orICU course may also influence the development of de-lirium.'"* These risk factors have been validated in thepresent study. Our results show that the risk of deliriumis significantly increased in elderly patients, those withprevious comorbidities, those admitted for trauma ofsignificant severity, as well as those involving specifi-cally a fall or bum and requiring ICU admission.

Although our study has the largest sample size ofany delirium study presently in the literature, severallimitations must be addressed. The NTDB collects itsdata from those centers willing to contribute to thedatabase. Therefore, our variable of interest, delirium,is limited to those centers that are willing or capable ofreporting data to the NTDB. In this study, ICD-9 codesfor delirium were used exclusively as a diagnostic tool.The present dataset did not contain details on pre-sentation or clinical symptomatology of delirium.Additionally, ethanol levels were lacking in approxi-mately 50 per cent of patients. Furthermore, for thosetesting positive for ethanol on admission, the quantity,duration, and frequency of ethanol ingestion could notbe extracted. The magnitude of the ethanol levels inparticular, would have allowed for the determination ofa threshold after which the risk of developing deliriumincreases. In addition, we could not accurately extractdata on the rates of sepsis or infections and narcotic orsedafive use in the present populafion. These are im-portant factors that have been shown to significantlyaffect the diagnosis of delirium. Finally, the timingwhen patients developed delirium as well as thetreatment received was not available for analysis. Thiswould have allowed for further insight into whetherdelirium is a cause or consequence of an increasedlength of stay, ICU LOS, or ventilation days.

When one adds the errors inherent to a retrospectiveexamination to an entity with a low incidence, a signif-icant limitafion to this study is the diagnosis of delirium.Even prospectively, it may be difficult to accuratelydiagnose delirium in a trauma patient who may havenumerous confounding factors including head injury,medications, electrolyte or metabolic derangements.

and infection. Despite these limitations, this study pro-vides epidemiologic data on the risk factors for thedevelopment of delirium in adult patients assessed attrauma centers across the United States.

In summary, the incidence of delirium in this traumapatient cohort was 0.6 per cent. Nine independent riskfactors for the development of delirium were identi-fied: 1) lack of insurance, 2) presence of ethanol onadmission, 3) chronic ethanol use, 4) ICU admission,5) age ^ 55 years, 6) burns, 7) Medicare insurance, 8)falls, and 9) previous history of cardiovascular disease.This data may be helpful in designing interventions toprevent delirium.

REFERENCES

1. Spies CD, Nordmann A, Brummer G, et al. Intensive careunit stay is prolonged in chronic alcoholic men following tumorresection of the upper digestive tract. Acta Anaesthesiol Scand1996;40:649-56.

2. Spies CD, Neuner B, Neumann T, et al. Intereurrent eom-plications in chronic alcoholic men admitted to the intensive careunit following trauma. Intensive Care Med 1996;22:286-93.

3. Marik P, Mohedin B. Alcohol-related admissions to an innercity hospital intensive eare unit. Alcohol Alcohol 1996;31:393-6.

4. Spies CD, Dubisz N, Neumann T, et al. Therapy of alcoholwithdrawal syndrome in intensive care unit patients followingtrauma: results of a prospective, randomized trial. Crit Care Med1996;24:414-22.

5. Iacobone E, Bailly-Salin J, Polito A, et al. Sepsis-associatedencephalopathy and its differential diagnosis. Crit Care Med 2009;37(Suppl):S331-6.

6. Ramehand R, Marshall GN, Schell TL, et al. Alcohol abuseand illegal drug use among Los Angeles County trauma patients:prevalenee and evaluation of single item sereener. J Trauma 2009;66:1461-7.

7. Lukan JK, Reed DN Jr, Looney SW, et al. Risk factors fordelirium tremens in trauma patients. J Trauma 2002;53:901-6.

8. Blondell RD, Powell GE, Dodds HN, et al. Admissioneharacteristies of trauma patients in whom delirium develops. Am JSurg2004;t87:332-7.

9. Linn LS, Gelberg L, Leake B. Substance abuse and mentalhealth status of homeless and domieiled low-ineome users ofa medical clinic. Hosp Community Psychiatry 1990;41:306-10.

10. Stein JA, Gelberg L. Comparability and representativenessof clinical homeless, community homeless, and domiciled clinicsamples: Physical and mental health, substance use, and healthservices utilization. Health Psychol 1997;16:155-62.

11. Rosen H, Saleh F, Lipsitz S, et al. Downwardly mobile: theaccidental cost of being uninsured. Arch Surg 2009;144:1006-11.

12. Salim A, Ottochian M, DuBose J, et al. Does insurancestatus matter at a public, level I trauma center? J Trauma 2010;68:211-6.

13. Isbell H, Fraser HF, Wilkler A, et al. An experimental studyof the etiology of rum fits and delirium tremens. Q J Stud Aleohol1955;16:l-33.

14. Goreliek DA, Wilkins JN. Special aspects of human aleoholwithdrawal. Recent Dev Alcohol 1986;4:283-305.

Copyright of American Surgeon is the property of Southeastern Surgical Congress and its content may not be

copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.