ÌÇÐÄvlog

ContextÌý Although posttraumatic stress disorder (PTSD) and alcohol abuse frequently occur among acutely injured trauma survivors, few real-world interventions have targeted these disorders.

ObjectiveÌý We tested the effectiveness of a multifaceted collaborative care (CC) intervention for PTSD and alcohol abuse.

DesignÌý Randomized effectiveness trial.

ParticipantsÌý We recruited a population-based sample of 120 male and female injured surgical inpatients 18 or older at a level I trauma center.

InterventionÌý Patients were randomly assigned to the CC intervention (n = 59) or the usual care (UC) control condition (n = 61). The CC patients received stepped care that consisted of (1) continuous postinjury case management, (2) motivational interviews targeting alcohol abuse/dependence, and (3) evidence-based pharmacotherapy and/or cognitive behavioral therapy for patients with persistent PTSD at 3 months after injury.

Main Outcome MeasuresÌý We used the PTSD symptomatic criteria (PTSD Checklist) at baseline and 1, 3, 6, and 12 months after injury, and alcohol abuse/dependence (Composite International Diagnostic Interview) at baseline and 6 and 12 months after injury.

ResultsÌý Random-coefficient regression analyses demonstrated that over time, CC patients were significantly less symptomatic compared with UC patients with regard to PTSD (P = .01) and alcohol abuse/dependence (P = .048). The CC group demonstrated no difference (−0.07%; 95% confidence interval [CI], −4.2% to 4.3%) in the adjusted rates of change in PTSD from baseline to 12 months, whereas the UC group had a 6% increase (95% CI, 3.1%-9.3%) during the year. The CC group showed on average a decrease in the rate of alcohol abuse/dependence of −24.2% (95% CI, −19.9% to −28.6%), whereas the UC group had on average a 12.9% increase (95% CI, 8.2%-17.7%) during the year.

ConclusionsÌý Early mental health care interventions can be feasibly and effectively delivered from trauma centers. Future investigations that refine routine acute care treatment procedures may improve the quality of mental health care for Americans injured in the wake of individual and mass trauma.

Injured survivors of individual and mass trauma receive their initialtreatment in acute care settings.¹ Within 48hours after the September 11, 2001, attack on the World Trade Center, 1103physically injured survivors were triaged through 5 acute care facilitiesin Manhattan, NY.² Each year approximately2.5 million Americans are so severely injured that they require inpatienthospital admission.³

Symptoms consistent with a diagnosis of posttraumatic stress disorder(PTSD) may develop in 10% to 40% of hospitalized injured patients in the UnitedStates.^4-7 Approximately20% to 40% of injured patients admitted to trauma centers have current orlifetime alcohol abuse/dependence diagnoses.⁸ Alcoholintoxication at the time of injury is associated with an increased risk ofinjury recurrence.⁹

Efficacy research suggests that individuals with PTSD respond to psychotherapeuticand psychopharmacological treatments.^10-12 Growingrandomized clinical trial evidence suggests that early cognitive behaviortherapy (CBT) interventions delivered in the days and weeks after injury canhelp to diminish PTSD symptom development.^13-16 Selectiveserotonin reuptake inhibitors and tricyclic antidepressants are efficacioustreatments for PTSD.^11,17-21 Efficacystudies suggest that motivational interviewing (MI) interventions can decreasealcohol use across a variety of clinical populations,^22-25 includinginjured trauma survivors.²⁶

Recent consensus guidelines from the National Institute of Mental Healthidentify acutely injured trauma survivors as a group at high risk for developmentof PTSD and related comorbid conditions and recommend the development of earlyevaluation procedures that are adaptable to real-world treatment settings.²⁷ As is true for many Americans with psychiatric disorders,²⁸ injured patients appear to receive fragmented care,and most are not engaged in mental health services at strategic postinjurypoints.^29,30 Previous investigationsof psychological debriefing suggest that although this intervention may befeasibly delivered to representative samples of patients receiving acute care,^31,32 debriefing interventions are noteffective in reducing PTSD symptoms and may actually be associated with pooreroutcomes.^33,34

During the past decade, collaborative care (CC) has been developed asa comprehensive treatment delivery model for patients with medical and psychiatricdisorders.^35-43 Collaborativecare is a disease management strategy that uses multifaceted interventions(eg, combined case management, pharmacotherapy, and psychotherapy) with theaim of integrating mental health interventions into general medical care.Just as collaborative models have improved mental health outcomes for patientswith major depression and panic disorder in primary care, the introductionof CC interventions within trauma care systems may link acute injury carewith evidence-based mental health interventions. Collaborative interventionshave the potential to reduce posttraumatic symptoms and trauma recidivismwhile improving functional recovery.

We developed and tested a multifaceted CC intervention targeting PTSDand alcohol use for acutely injured trauma survivors. The primary hypothesiswas that patients receiving the CC intervention would demonstrate significantreductions in PTSD and alcohol abuse during the year after injury.

Subjects were recruited from the Harborview level I trauma center ofthe University of Washington, Seattle. Each year Harborview admits approximately5000 survivors of intentional (eg, injuries associated with human malice suchas physical assaults) and unintentional (eg, injuries associated with motorvehicle crashes and job-related injuries) injuries of all ages. Eligible patientswere English-speaking survivors of intentional and unintentional injuries,18 years and older, who lived within 50 miles of the trauma center. Harborviewtrauma registry data documenting injury, demographic, and clinical variableswere available for screening each patient admitted during the study period.The University of Washington institutional review board approved all traumaregistry analyses and informed consent procedures before the initiation ofthe study.

On weekdays from March 30, 2001, through January 10, 2002, a researchassociate downloaded an automated list of all injured patients admitted foracute injuries to Harborview's trauma surgical services. Eligible, newly admittedpatients were approached using random number assignments from a computer-generatedalgorithm. With regard to cognitive status, patients approached in the wardwere required to have a Glasgow Coma Scale score⁴⁴ of15 and a score of at least 7 on the 2 Mini-Mental State Examination itemsthat assess orientation to location and date.⁴⁵ Patientswith severe injuries that prevented participation were excluded from the study.Patients who had self-inflicted injuries or active psychosis, who were currentlyincarcerated, or who had recent histories of violence were also excluded.

Hospitalized inpatients received a 2-phased evaluation. First, we administeredthe PTSD Checklist Civilian Version (PCL)⁴⁶ andthe Center for Epidemiological Studies Depression Scale (CES-D)⁴⁷ toeach inpatient. We included patients in the study who were symptomatic withPTSD (PCL score, ≥45)⁴⁸ and/or depression(CESD-D score, ≥16).⁴⁹ Second, patientswho met symptomatic criteria for the study were administered the remainderof the surgical ward interview that consisted of the Composite InternationalDiagnostic Interview (CIDI) alcohol abuse/dependence modules⁵⁰ andother items assessing various aspects of preevent- and event-related symptoms,functioning, and use of health care services.

Patients were randomized to the CC intervention or the usual care (UC)control condition immediately after completion of the surgical ward assessment.Randomization was stratified according to PTSD symptom levels (PCL score,≥45 vs <45), depressive symptom levels (CES-D score, ≥16 vs <16),and results of alcohol and stimulant admission toxicology (positive vs negative).

The CC intervention combined case management and psychopharmacologicaland psychotherapeutic treatments. The intervention team included a full-timemaster's level case manager, the trauma support specialist (TSS) (R.D.), andpart-time psychiatrist (D.Z.) and psychologist (A.W.) interventionists whodelivered the medication and CBT components. The intervention procedure hasbeen manualized, and the component modules have been published.^51-53

The intervention was designed as a stepped-care procedure. For the first6 months after injury, all CC patients received continuous case managementdelivered by the TSS. The TSS was the frontline provider in the treatmentof the injured CC patient and provided readily accessible, continuous traumasupport in the weeks and months after the injury.

The TSS began treatment with each injured CC patient at the bedsidein the surgical ward. Prior investigation suggested that case management facilitatedengagement in mental health intervention among ethnically diverse low-incomepatients.⁵⁴ In previous studies, injured traumasurvivors demonstrated multiple posttraumatic concerns (eg, physical health,work, and finance) that extended beyond PTSD and alcohol-related symptoms.^55,56 Thus, to engage injured patients,the TSS elicited, tracked, and targeted for improvement each injured patient'sunique constellation of posttraumatic concerns.

The TSS coordinated care across surgical inpatient, primary care outpatient,specialty mental health, and community service settings. With the other membersof the intervention team, the TSS developed a comprehensive postinjury careplan that simultaneously addressed the medical and psychosocial complicationsof the injury and coordinated linkages to primary care and community services.In these activities, the TSS interfaced with patients and their families,surgical and primary care providers, staff at community agencies, and outsidemental health care professionals. The case management pager was covered bymembers of the intervention team 24 hours a day, 7 days a week, to providecare that was responsive to the spontaneous questions and needs of injuredpatients.⁵⁷ These combined trauma support activitiesestablished a therapeutic alliance that facilitated the delivery of evidence-basedinterventions for alcohol abuse and PTSD.

All patients with positive alcohol toxicology test results on admission,or who at any point during the trial demonstrated postinjury alcohol abusethat could be considered hazardous and risked injury recurrence, receivedthe evidence-based MI intervention.^26,52,58 TheTSS was trained by the team's expert MI consultant (C.D.) to deliver MI targetingalcohol abuse in trauma wards.⁵⁹ The MI interventionconsisted of an initial surgical ward session followed by as-needed boostersessions. The surgical ward session lasted approximately 30 minutes, and thefollow-up booster sessions had variable lengths ranging from 10 to 60 minutes.The intervention components included feedback about inpatient blood alcoholtoxicology test results, exploration of the pros and cons of alcohol consumption,discussion of the importance of change, clarification of specific goals foralcohol use, and negotiation of action plans to bring about change. Any patientwho requested MI booster sessions received them; patients who had exacerbationsof alcohol use were also offered booster sessions.

Patients who demonstrated high levels of immediate posttraumatic distress(eg, severe anxiety, pain, and/or insomnia) received early psychiatric evaluations.⁵³ Because immediate posttraumatic distress can spontaneouslyresolve in the weeks and months after injury, only patients with sustainedhigh levels of early distress were offered evidence-based PTSD pharmacologicalinterventions (eg, selective serotonin reuptake inhibitors) in the first 3months after the event. Sustained high levels of distress were operationalizedas (1) objectively observed high levels of distress such as extreme emotionalreactions (eg, fear, rage, and dissociation) that lasted at least 24 hoursand were so severe as to limit verbal interchanges and/or (2) sustained subjectivedistress lasting days that prompted repetitive patient requests for more intensivetreatment. Based on these criteria, 4 (7%) of the 59 CC patients receivedPTSD pharmacological intervention before the 3-month postinjury time point.

Three months after the injury, each CC patient was administered theStructured Clinical Interview for DSM-IV PTSD moduleby the case manager.⁶⁰ All patients who receiveda PTSD diagnosis at this assessment were given their preference of CBT, pharmacotherapy,or combined treatment. Evidence-based PTSD treatments were delivered by theteam's expert psychotherapy and pharmacotherapy consultants. The CBT interventionincluded psychoeducation, muscle relaxation, cognitive restructuring, andgraded exposure.^13,51 The psychopharmacologicalintervention consisted of an initial psychiatric evaluation and medicationtargeting PTSD.^61,62

When care for PTSD was stepped up at the 3-month postinjury point, theTSS provided education about the diagnosis and facilitated the entry of patientsinto evidence-based treatments. During the evidence-based PTSD intervention,the TSS had the key role of performing brief assessments of adherence to medicationtherapy and symptom relapse, outside scheduled psychotherapy or medicationsessions.

The stepped-care procedure included relapse prevention and communityintegration components. From 6 to 12 months after the injury, all patientswho remained symptomatic with PTSD and/or demonstrated evidence of alcoholabuse/dependence or alcohol consumption behaviors that risked injury recurrencereceived combinations of ongoing trauma support and evidence-based MI andPTSD treatments. In this phase of the protocol, the TSS remained in contactwith the patient and periodically reassessed symptoms, function, and rehabilitation.

The collaborative team members maintained detailed logs documentingthe nature and duration of all intervention activities.⁵⁶ Thecollaborative team held weekly meetings to review cases and protocol procedures.

Patients assigned to the control condition received care as usual. Previousinvestigations have documented that injured trauma survivors typically receivecare from a heterogeneous group of clinicians including surgical practitioners,emergency department caregivers, and primary care providers.⁶³ Acuteposttraumatic distress is infrequently detected and treated in the surgicalinpatient setting,⁶⁴ and less than 15% of injuredtrauma survivors report receiving specialty mental health and/or substance-relatedcare during the year after injury.^52,63 Inthis investigation all patients received a list of community referrals immediatelyafter their surgical ward assessment; 21% of UC controls (11 of the 53 withself-report health service utilization data) endorsed 1 or more visits withmental health specialty providers (ie, at the doctoral level) during the courseof the year after injury.

The principal investigator, a consultation liaison psychiatrist (D.Z.),oversaw the training procedure for the surgical ward and telephone follow-upinterviews. A research associate made morning rounds with the principal investigatorduring an initial 1-month pilot phase and then for the following 3 monthsof active protocol recruitment. During this period all recruitment, consent,and interview procedures were observed and critiqued by the principal investigator.Training of research associates for structured clinical assessments and telephoneinterviews included the use of practice interviews and manuals. These interviewswere monitored for reliability by the principal investigator and senior coinvestigators.

Symptoms consistent with a diagnosis of PTSD were assessed with thePCL, a 17-item self-report questionnaire that elicits graded responses (range,1-5) for the intrusive, avoidant, and arousal PTSD symptom clusters.⁴⁶ A series of investigations have demonstrated thereliability and validity of the PCL across trauma-exposed populations.^46,48,65-68

Responses are recorded on a scale from "not at all" (grade 1) through"moderately" (grade 3) to "extremely" (grade 5). Symptoms consistent withthe DSM-IV diagnosis were determined by adherenceto the recommended algorithm that considers a score of 3 or greater a symptomand follows the diagnostic rules requiring at least 1 intrusive symptom, 3avoidant symptoms, and 2 arousal symptoms.^46,69 Symptomsof PTSD were assessed in reference to the index injury (eg, "How botheredhave you been by repeated, disturbing memories, thoughts, or images of theevent in which you were injured?"). We used the PCL to screen for symptomsin the surgical ward and to reassess for PTSD at the 1-, 3-, 6-, and 12-monthtelephone follow-up interviews.

The DSM-IV diagnoses of alcohol abuse and dependencewere assessed with the CIDI Alcohol module.⁷⁰ TheCIDI, a structured diagnostic interview developed by the World Health Organizationand the US Alcohol, Drug Abuse, and Mental Health Administration, has establishedreliability and validity for the DSM diagnoses ofalcohol abuse/dependence.⁷⁰ The CIDI alcoholabuse/dependence modules were administered at baseline in the surgical wardand again 6 and 12 months after the injury. The baseline CIDI assessed alcoholconsumption behaviors during the 12 months before the injury, whereas the6- and 12-month telephone follow-up CIDI evaluations assessed each preceding6-month period.

Alcohol and stimulant (amphetamine and cocaine) intoxication at thetime of admission to the hospital was assessed with toxicology screens. Becauseopiates and benzodiazapines are frequently administered to trauma patients,only stimulant results were included as positive results of drug screens.

We determined injury severity from the medical record International Classification of Disease, Ninth Version, Clinical Modification (ICD-9-CM)⁷¹ codesusing the Abbreviated Injury Scale and Injury Severity Score.⁷² Comorbidchronic medical conditions were also derived from ICD-9-CM diagnostic codes. Eleven conditions, including diabetes, hypertension,chronic liver and renal diseases, carcinomas, ischemic heart disease, degenerativenervous conditions, stroke, epilepsy, obesity, and coagulation defects wereincluded.⁷³

To assess the representativeness of the study sample, we first comparedthe characteristics of patients included in the investigation with the characteristicsof all injured patients admitted to Harborview during the study period whomet study eligibility criteria. We also compared baseline data for patientsin the CC and UC conditions.

We used mixed-effects random-coefficient regression models with theintent-to-treat sample to determine whether patients in the 2 groups manifesteddifferent patterns of PTSD and alcohol symptom change over time. Longitudinaldata collected prospectively from injured trauma survivors is characterizedby correlated intraindividual observations, missing data, and dropouts.⁴ Mixed-effects random-coefficient regression methodswere selected because of their superior ability to model longitudinal datawith these characteristics.⁷⁴ The procedureuses maximum likelihood estimates to evaluate treatment group, time, and treatmentgroup × time interaction effects. Baseline symptom status, age, sex,medical comorbidity, and injury severity were used as covariates in all theseregression procedures.

When significant interaction or main effects were detected by the random-coefficientregressions, post hoc logistic regressions for each follow-up assessment pointwere performed. On the basis of these logistic regressions, we calculatedthe average rates of PTSD and alcohol abuse/dependence at each assessmentpoint for the CC and UC groups adjusted for the covariates. These adjustedaverage proportions are presented in our figures. To interpret any significantstatistical interactions of treatment with time, we used logistic regressionsto calculate adjusted probabilities of PTSD and alcohol abuse/dependence foreach patient at each assessment. Next, changes in the adjusted probabilitiesfrom baseline to the 6- and 12-month assessments were calculated along withtheir 95% confidence intervals (CIs). Finally, we used 2-tailed unpaired t tests to examine the differential rates of change inthe adjusted probabilities across time for the treatment groups. Significantdifferences in the CC and UC groups would indicate differences in the ratesof change in adjusted probabilities over time.

A total of 2610 surgical inpatients underwent screening for the investigationduring the 10-month study (Figure 1).Injured trauma survivors recruited into the investigation did not significantlydiffer from all patients admitted to Harborview surgical services during thestudy period with regard to injury type (intentional injury, 21% vs 19%; χ²₁ = 0.18; P = .67), Injury SeverityScore (mean, 10.8 [SD, 6.6] vs 10.8 [SD, 9.4]; t₂₁₅₇ = 0.01; P = .99), sex (percentage female,33% vs 27%; χ²₁ = 1.2; P =.27), or alcohol toxicology test status (percentage with positive results,28% vs 28%; χ²₂ = 1.88; P =.39). On average, study patients were younger (mean age, 38.7 years [SD, 14.8years] vs 41.7 years [SD, 18.2 years]; t₂₅₀₉ = 0.01; P = .07) and significantly less likelyto have 1 or more chronic medical conditions (14% vs 26%; χ²₁ = 6.9; P = .01) when compared with the populationof trauma center patients.

Injured patients included in the investigation (n = 120) were heterogeneouswith regard to demographic and clinical characteristics (Table 1). Sixty-six percent of patients were white; 12%, AfricanAmerican; 10%, Hispanic; 8%, Native American; and 5%, Asian. Thirty-six percentof patients had individual incomes less than $15Ìý000 per year; 34%, from$15Ìý000 to $40Ìý000 per year; and 30%, greater than $40Ìý000per year. Eleven percent of patients reported not having a permanent homeor living situation. Less than 10% of patients had positive findings for stimulantson admission, and the number of patients with stimulant-positive results wasnot significantly different across the 2 groups. Twelve (48%) of the 25 patientswho met DSM-IV symptomatic criteria for PTSD in thesurgical ward had comorbid alcohol abuse/dependence. Although the proportionsof CC and UC patients with positive alcohol admission toxicologic findingswere similar, stratification based on this crude screen failed to evenly distributepatients with regard to alcohol abuse/dependence (Table 1).

The stepped-care procedure involved gradually decreasing case managertime intensity during the weeks and months after the injury and graduallyincreasing time commitments from doctoral-level practitioners. The case managerbegan each intervention with a bedside visit, and during the course of theyear spent an average of 10.7 hours (SD, 9.8 hours; median, 7.6 hours; range,1-56 hours) with each CC patient. On average, the case manager spent 4.3 hours(SD, 2.9 hours) with each patient in the first month after the injury, 3.1hours (SD, 3.65 hours) per patient from months 1 through 3, 2.0 hours (SD,3.4 hours) per patient from months 4 to 6, and 1.3 hours (SD, 3.2 hours) perpatient during months 7 to 12 after injury. Thirty (51%) of 59 CC patientsreceived the brief MI intervention targeting alcohol abuse, and more thanhalf of these patients received 1 or more MI booster sessions during the year.

At the 3-month assessment with the Structured Clinical Interview for DSM-IV, 12 (24%) of 50 CC patients received a diagnosisof PTSD. Evidence-based pharmacotherapy and psychotherapy were offered tothese patients. Almost all of the intervention time (38.9 [95.8%] of 40.6total hours) of the expert doctoral-level therapist was spent delivering acourse of CBT (range, 5-12 sessions) to 5 CC patients from 3 to 12 monthsafter injury.

In total, the psychiatrist participated in the evaluation and/or treatmentof 38 (64%) of the 59 CC patients (average time per patient, 2.7 hours; SD,3.4 hours; median, 0.67 hours; range, 0.1-14.5 hours). The psychiatrist participatedin an average of 0.60 visits of 30 to 60 minutes' duration in the first 3months (SD, 1.00; range, 0-5), and 1.0 visits from months 3 to 12 (SD, 2.1;range, 0-8). With regard to patient-related telephone calls, in the first3 months the psychiatrist averaged 1.0 telephone contacts per patient (SD,1.9; range, 0-9 calls), and from 3 to 12 months, 1.4 telephone contacts (SD,2.3; range, 0-9). In the first 3 months immediately after the injury, thepsychiatrist evaluated 22 (37%) of 59 CC patients for high levels of immediateposttraumatic distress, pain, insomnia, or other injury-related complications.Also, for 4 (7%) of 59 patients, no in-person evaluation was performed; however,the psychiatrist participated in care coordination/referral activities withsurgical and primary care providers. Twenty (34%) of the 59 CC patients wereoffered PTSD pharmacotherapy; 10 (50%) of these 20 patients accepted and maintainedtheir medication regimes.

The random regression procedure showed a significant treatment group× time interaction effect for PTSD (Table 2). The intervention effect coincided with the initiationof evidence-based medication and psychotherapy interventions at 3 months (Figure 2). The significant treatment group× time interaction was due to treatment group differences in the adjustedrates of change in PTSD for the CC and UC groups. At 6 months, the CC andUC groups had trend level differences in rates of change from baseline (t₁₁₈ = 1.83; P = .07).The CC group had on average a 5.5% increase in the rate of PTSD (95% CI, 0.1%-10.8%),whereas the UC group had on average twice the rate of increase in the first6 months (12.0%; 95% CI, 7.3%-16.7%). However, at 12 months, the differencesin rates of PTSD from baseline were statistically different (t₁₁₈ = 2.40; P = .02), with theCC group showing no change in the rate of PTSD (a decrease of 0.07%; 95% CI,−4.2%to 4.3%) and the UC group showing on average a 6% increase in the rate ofPTSD during the year (95% CI, 3.1%-9.3%).

A significant treatment group × time interaction effect was observedfor CIDI-diagnosed alcohol abuse/dependence (Table 2). The intervention appears to have maintained reductionsof alcohol consumption beyond 6 months (Figure3). The significant treatment group × time interaction wasdue to treatment group differences in the adjusted rates of change in alcoholabuse/dependence for the CC and UC groups. At 6 months, the CC and UC groupshad significantly different rates of change from baseline (t₁₁₈ = 3.37; P = .001). The CCgroup had on average a 20.4% decrease in the rate of alcohol abuse/dependence(95% CI, −14.3% to −26.5%), whereas the UC group had on averageonly a 7% decrease (95% CI, −2.8% to −12.2%). At 12 months, thedifferences in rates of alcohol abuse/dependence were also statistically different(t₁₁₈ = 11.53; P<.001),with the CC group showing on average a decrease in the rate of alcohol abuse/dependenceof 24.2% during the year (95% CI, −19.9% to −28.6%) and the UCgroup showing on average a 12.9% increase in the rate of alcohol abuse/dependenceduring the year (95% CI, 8.2%-17.7%).

This investigation establishes the feasibility of delivering a multifacetedCC intervention to acutely injured trauma survivors. We successfully recruited,intervened with, and followed up a representative sample of injured traumasurvivors, some of whom had alcohol abuse/dependence and acute posttraumaticdistress. Early case management intervention successfully engaged acutelytraumatized patients, a population that in the past has demonstrated a reluctanceto participate in ongoing mental health interventions.^76-78 TheCC intervention differed from previous trials of debriefing in that earlycase management activities established an ongoing relationship with patients,who were later offered evidence-based treatment.

The investigation demonstrated that a stepped CC delivery model effectivelyreduces alcohol abuse/dependence during the year after injury. At 6 and 12months after the injury, clinically and statistically significant reductionsin alcohol use were apparent for patients who received the CC intervention.

For PTSD, CC patients demonstrated essentially no change in symptomsduring the course of the year, whereas UC patients manifested a significantworsening of symptoms. Early evaluation and supportive intervention was notassociated with reductions in PTSD for patients in the CC condition. Preventionof the development of PTSD in CC patients relative to UC patients coincidedwith the initiation of evidence-based PTSD medication and CBT treatments 3months after injury.

There are a number of possible explanations for the CC intervention'srelatively small PTSD treatment effect. The investigation was limited by aninitial screening procedure that recruited some patients with minimal PTSDsymptoms. Also, previous investigations of injured trauma survivors hospitalizedat level I trauma centers in the United States suggest that this patient populationexperiences multiple recurrent traumatic life events, including a substantialburden of traumatic injuries requiring hospitalization, some of which followan index injury admission.^4,9,26,79-81 Ourinvestigation is limited in that we did not specifically assess PTSD in relationto these multiple prior and subsequent traumatic life events. It may be thatthe CC intervention buffers CC patients from the full symptomatic impact ofrecurrent traumatic life events, relative to UC patients. These preliminaryobservations will require more refined study in future investigations.

There are other important considerations in interpreting the resultsof this investigation. The design of this study builds on a series of effectivenesstrials for depressive and anxiety disorders delivered in real-world treatmentsettings.^37-42,82 Thetrade-offs relevant to the effectiveness approach apply to the current investigation.⁸³ Because this was a multifaceted intervention, itdid not yield information regarding which components of the treatment areefficacious. For instance, although PTSD prevention was temporally associatedwith the initiation of evidence-based treatments, we cannot rule out the possibilitythat continuing trauma support activities also contributed to the preventionof PTSD. Finally, as is typical of many effectiveness trials, we relied onsymptom screens and lay interviews rather than clinician-administered diagnosticassessments.

This investigation contributes to a developing literature regardingearly intervention for posttraumatic distress in acutely traumatized patients.A stepped CC intervention furthers the delivery of high-quality posttraumaticcare by tailoring treatment needs to the individual trauma survivor whiledelivering evidence-based mental health interventions.⁵⁷ Collaborativecare is a multifaceted disease management strategy into which future psychotherapeuticand pharmacological advances for the treatment of PTSD can be incorporated.^84,85 Future larger-scale CC trials shouldassess functional outcome improvements and the cost-effectiveness of the intervention.⁸⁶

The September 11, 2001, attack on American civilians provides an additionalimpetus for the ongoing development of multifaceted acute care mental healthscreening and intervention procedures. Injured trauma survivors triaged throughacute care in the immediate aftermath of a mass attack represent a high-risksubgroup of patients who are rapidly transported to central points of contactwithin the health care system.⁸⁷ Commentariesthat have followed the September 11 attack suggest that the health care systemin the United States should be better prepared for mass civilian trauma.^27,88 Future investigations that refineroutine acute care evaluation and treatment procedures have the potentialto improve the quality of mental health care for Americans injured in thewake of individual and mass trauma.

Corresponding author: Douglas Zatzick, MD, University of WashingtonSchool of Medicine, Box 359911, Harborview Medical Center, 325 Ninth Ave,Seattle WA 98104 (e-mail: dzatzick@u.washington.edu)

Submitted for publication August 15, 2003; final revision received December15, 2003; accepted January 20, 2004.

This study was supported by grant 1K08 MH01610 from the National Instituteof Mental Health, Rockville, Md; grant 1R03 HS11372 from the Agency for HealthcareResearch and Quality, Rockville; and grant CCR303568 from the Centers forDisease Control and Prevention, Atlanta, Ga.

A preliminary version of this study was presented at the Annual Meetingof the Robert Wood Johnson Clinical Scholars Program; November 22, 2003; FtLauderdale, Fla.