STAR*D: Results and STAR*D: Results and Implications for Clinicians, Implications for Clinicians, Researchers, and Policy Researchers, and Policy
MakersMakers
Bradley N. Gaynes, M.D., M.P.H.Bradley N. Gaynes, M.D., M.P.H.Associate Professor of PsychiatryAssociate Professor of Psychiatry
University of North Carolina School of University of North Carolina School of MedicineMedicine
Chapel Hill, North CarolinaChapel Hill, North Carolina
AcademyHealth Annual Research Meeting AcademyHealth Annual Research Meeting 20072007
What is STAR*D?What is STAR*D?
Sequenced Treatment Alternatives to Sequenced Treatment Alternatives to Relieve DepressionRelieve Depression
www.star-d.orgwww.star-d.org
Overall Aim of STAR*DOverall Aim of STAR*D
Define preferred Define preferred treatments for treatment-treatments for treatment-resistant depressionresistant depression
44
Overview - IOverview - I Duration: 7 years (October 1999 - Duration: 7 years (October 1999 -
September 2006)September 2006) Funding: National Institute of Mental Funding: National Institute of Mental
HealthHealth National Coordinating Center, National Coordinating Center,
UT Southwestern Medical Center, UT Southwestern Medical Center, DallasDallas
Data Coordinating Center, PittsburghData Coordinating Center, Pittsburgh
Overview - IIOverview - II
14 Regional Centers14 Regional Centers 41 Clinical Sites41 Clinical Sites 18 Primary Care Settings (PC)18 Primary Care Settings (PC) 23 Psychiatric Care Settings 23 Psychiatric Care Settings
(Specialty Care, or SC)(Specialty Care, or SC)
Obtain ConsentObtain Consent
CITCIT Follow-UpFollow-Up
Level 2Level 2
Satisfactory responseSatisfactory response
Unsatisfactory response*Unsatisfactory response*
*Response = >50% improvement in QIDS-SR from baseline
*Response = >50% improvement in QIDS-SR from baseline
Level 1Level 1
Level 2Level 2
RandomizeRandomizeRandomizeRandomize
Switch OptionsSwitch Options Augmentation OptionsAugmentation Options
SERSER BUP-SRBUP-SR VEN-XRVEN-XR CTCTCIT +
BUP-SRCIT +
BUP-SRCIT + BUS
CIT + BUS
CIT + CT
CIT + CT
Level 2ALevel 2A
RandomizeRandomizeRandomizeRandomize
Switch OptionsSwitch Options
BUP-SRBUP-SR VEN-XRVEN-XR
Level 3Level 3
RandomizeRandomizeRandomizeRandomize
Switch OptionsSwitch Options Augmentation OptionsAugmentation Options
MRTMRT NTPNTPL-2 Tx
+ LiL-2 Tx
+ LiL-2 Tx+ THYL-2 Tx+ THY
Level 4Level 4
RandomizeRandomizeRandomizeRandomize
Switch OptionsSwitch Options
TCPTCPVEN-XR+ MRT
VEN-XR+ MRT
ParticipantsParticipants Major depressive disorderMajor depressive disorder NonpsychoticNonpsychotic Representative primary and Representative primary and
specialty care practices specialty care practices (nonacademic/non efficacy venues)(nonacademic/non efficacy venues)
Self-declared patientsSelf-declared patients
Inclusion CriteriaInclusion Criteria Clinician deems antidepressant Clinician deems antidepressant
medication indicated.medication indicated.
18-75 years of age.18-75 years of age.
Baseline HRSDBaseline HRSD17 17 14.14.
Most concurrent Axis I, II, III Most concurrent Axis I, II, III disorders allowed.disorders allowed.
Suicidal patients allowedSuicidal patients allowed
Clinical ProceduresClinical Procedures Open treatment with randomizationOpen treatment with randomization
Symptoms/side effects measured at Symptoms/side effects measured at each clinical visit (measurement-each clinical visit (measurement-based care, or MBC)based care, or MBC)
Clinicians guided by algorithms/Clinicians guided by algorithms/supervisionsupervision
Research InnovationsResearch Innovations ““Real world” patient participants from Real world” patient participants from
nonacademic/nonefficacy research nonacademic/nonefficacy research venues venues
Non-research clinicians Non-research clinicians Identical criteria and concurrent Identical criteria and concurrent
enrollment from PC and SC sitesenrollment from PC and SC sites Broadly selective inclusion criteriaBroadly selective inclusion criteria Patient preference built into study designPatient preference built into study design
STAR*D Hybrid Design - ISTAR*D Hybrid Design - IEfficacy*Efficacy* EffectivenesEffectivenes
ssSTARSTARDD
PatientsPatients SymptomatSymptomatic ic
VolunteersVolunteers
Self-declared Self-declared Self-Self-declareddeclared
Masked Masked TreatmentTreatment
YesYes NoNo NoNo
Masked RatersMasked Raters YesYes YesYes YesYesBaseline Baseline SeveritySeverity
HRSDHRSD1717 >>2020
VariableVariable HRSDHRSD1717 >>1414
Diagnostic Diagnostic MethodMethod
Structured Structured InterviewInterview
ClinicalClinical ClinicalClinical
Concurrent Axis I Concurrent Axis I and Axis III and Axis III AllowedAllowed
MinimalMinimal Most†Most† Most†Most†
*To establish efficacy versus placebo.†Allowed to enter if MDD requires medication.
STAR*D Hybrid Design - IISTAR*D Hybrid Design - IIEfficacyEfficacy
**EffectivenesEffectivenes
ssSTARSTARDD
Treatment MethodsTreatment Methods ProtocolProtocol ClinicianClinician Protocol + Protocol + ClinicianClinician
Symptomatic Symptomatic OutcomesOutcomes
YesYes SometimesSometimes YesYes
Functional Functional OutcomesOutcomes
NoNo YesYes YesYes
Cost/Utilization Cost/Utilization OutcomesOutcomes
NoNo YesYes YesYes
Psychotherapy Psychotherapy AllowedAllowed
NoNo YesYes SometimeSometimes‡s‡
Placebo AllowedPlacebo Allowed YesYes NoNo NoNo
Suicidal Patients Suicidal Patients AllowedAllowed
NoNo YesYes YesYes*To establish efficacy versus placebo.‡Allowed if not depression-targeted, empirically tested therapy.
Level 1 FindingsLevel 1 Findings
Patients from real world Patients from real world settings are quite chronically illsettings are quite chronically ill
Mean (SD)
HRSD17 (ROA) 21.8 (5.2)
No. of MDEs 6.0 (11.4)
Length of current MDE (months) 24.6 (51.7)
Length of illness (years) 15.5 (13.2)
No.No. with either chronic or recurrent MDE 85%
Depressed ≥ 2 years 25%
No. with concurrent medical conditions 67%
Depressed patients in PC and Depressed patients in PC and SC settings are surprisingly SC settings are surprisingly
similar similar No difference in No difference in
depressive severitydepressive severity distribution of depressive severitydistribution of depressive severity specific depressive symptom presentationspecific depressive symptom presentation likelihood of presenting with a comorbid likelihood of presenting with a comorbid
psychiatric illnesspsychiatric illness Main difference: SC patients more Main difference: SC patients more
likely to have made prior suicide likely to have made prior suicide attempt, but common in both (20% vs. attempt, but common in both (20% vs. 14%, p<0.0001)14%, p<0.0001)
Outcomes for PC and SC Outcomes for PC and SC depressed patients were depressed patients were
identicalidentical Remission rates were the same (27% Remission rates were the same (27%
PC vs. 28% SC, p=0.40)PC vs. 28% SC, p=0.40)
Time to remission did not differ by Time to remission did not differ by site (6.7 weeks PC vs. 7.3 weeks CS, site (6.7 weeks PC vs. 7.3 weeks CS, p=0.11)p=0.11)
Gaynes et al., BMJ, under review
Time to Remission (QIDS-SR 16) by Clinical Setting
Log-Rank Test=2.6: p=0.1063
Weeks in Level 1
No. of patientsPrimary 1004 879 709 520 342 175 21Specialty 1643 1519 1254 975 633 294 28Total 2647 2398 1964 1495 975 469 49
Gaynes et al., BMJ, under review
ConclusionsConclusions One-quarter of patients have been One-quarter of patients have been
depressed for >2 years and 2/3 have depressed for >2 years and 2/3 have concurrent GMCsconcurrent GMCs
About 1/3 will remitAbout 1/3 will remit Response occurs in 1/3 AFTER 6 weeksResponse occurs in 1/3 AFTER 6 weeks MBC is feasible and works, with equivalent MBC is feasible and works, with equivalent
outcomes in PC or SC settingsoutcomes in PC or SC settings Studies of remission require longer study Studies of remission require longer study
periods than 8 weeksperiods than 8 weeks
Level 2 Medication SwitchLevel 2 Medication Switch
Conclusions: Level 2 SwitchConclusions: Level 2 Switch
Either switching to the same class of Either switching to the same class of antidepressant (SSRI to SSRI) or to a antidepressant (SSRI to SSRI) or to a different class (SSRI to non-SSRI) did different class (SSRI to non-SSRI) did not matternot matter
Substantial differences in pharmacology Substantial differences in pharmacology did not translate into substantial clinical did not translate into substantial clinical differences in efficacydifferences in efficacy
Level 2 Medication Level 2 Medication AugmentationAugmentation
Conclusions: Level 2 Conclusions: Level 2 AugmentationAugmentation
There was no substantial differences in There was no substantial differences in the likelihood of either of the two the likelihood of either of the two augmentation medications to produce augmentation medications to produce remissionremission
Patients had clear preferences about Patients had clear preferences about accepting augmentation vs. switching, accepting augmentation vs. switching, and, accordingly, the groups differed at and, accordingly, the groups differed at entry into level 2entry into level 2
Consequently, whether switching vs. Consequently, whether switching vs. augmenting is preferred after one augmenting is preferred after one treatment failure could not be treatment failure could not be addressedaddressed
QIDS-SRQIDS-SR16 16 Remission RatesRemission Rates
53.0%
30.6%32.9%
0
20
40
60
80
L-1 L-2 Overall
Percent
*53.0%
30.6%32.9%
0
20
40
60
80
L-1 L-2 Overall
Percent
*
* Theoretical* Theoretical
ConclusionsConclusions
Cumulative remission rate is over 50% Cumulative remission rate is over 50% with first 2 stepswith first 2 steps
Patient preference plays a big role in Patient preference plays a big role in strategy selectionstrategy selection
Pharmacological distinctions do not Pharmacological distinctions do not translate into large clinical differencestranslate into large clinical differences
Level 2 Cognitive Therapy Level 2 Cognitive Therapy FindingsFindings
ConclusionsConclusions CT is an acceptable switch option in CT is an acceptable switch option in
the second stepthe second step
CT is an acceptable augmentation CT is an acceptable augmentation option in the second stepoption in the second step
Whether CT responders/remitters Whether CT responders/remitters fare better in follow-up is in analysisfare better in follow-up is in analysis
CT was not as popular as expectedCT was not as popular as expected
Remission Rates by LevelsRemission Rates by Levelsaa
a By QIDS-SR16 <5 at level exit
Level 1 (2876) 32.9
Level 2 (1439) Switch (789) Augment (650)
30.627.035.0
Level 3 (377) Switch (235) Augment (142)
13.610.319.1
Level 4 (109) 14.7
Are Efficacy and Real Are Efficacy and Real World Patients Different?World Patients Different?
STAR*D Participant Flow STAR*D Participant Flow (CONSORT Chart) (CONSORT Chart)
Screened(4,790)
Not offered Consent
orRefused to
Consent(613)
Ineligible
(136)
Consented(4,177)
Efficacy Sample(635)
Nonefficacy Sample(2,220)
Could Not Be Classified
(21)
Failed to Return(234)
Eligible(4,041)
HRSD17 >14
(3,110)
Eligible for Analysis(2,876)
HRSD17 < 14a
(607)Or Missing
(324)
a Some of these subjects were eligible for entry into Level 2.Wisniewski et al, The Lancet, in preparation
Clinical FeaturesClinical Featuresaa
a Descriptive statistics presented as mean±sd and n (%N). Sums do not always equal N due to missing values. Percentages based on available data; b p<.01; c p<.05
Wisniewski et al, The Lancet, in preparation
FeatureFeatureEfficacyEfficacy(n=635)(n=635)
NonefficacNonefficacyy
(n=2220)(n=2220)Illness duration (yrs.)Illness duration (yrs.)bb 1313 1616Suicide attemptSuicide attemptcc 15%15% 19%19%Anxious featuresAnxious featuresbb 47%47% 55%55%Atypical featuresAtypical featuresbb 14%14% 20%20%Melancholic featuresMelancholic features 25%25% 23%23%Psychiatric carePsychiatric carebb 70%70% 59%59%
OutcomesOutcomesaa - I - I
a Descriptive statistics presented as mean±sd and n (%N). Sums do not always equal N due to missing values. Percentages based on available data
QIDS-SR16 = 16-item Quick Inventory of Depressive Symptomatology – Self-report
Wisniewski et al, The Lancet, in preparation
OutcomeOutcomeEfficacyEfficacy(n=635)(n=635)
NonefficacNonefficacyy
(n=2220)(n=2220)
QIDS-SRQIDS-SR1616 remission remission 35%35% 25%25%
QIDS-SRQIDS-SR1616 response response 52%52% 39%39%
Exit QIDS-SRExit QIDS-SR1616
8.68.6++5.25.210.010.0++5.65.6
QIDS-SRQIDS-SR1616 % change % change --45.445.4++33.233.2
--37.437.4++33.333.3
OutcomesOutcomesaa - II - II
a Descriptive statistics presented as mean±sd and n (%N). Sums do not always equal N due to missing values. Percentages based on available data; b Adjusted for regional center, clinical setting, age, race, Hispanic ethnicity, education, employment status, income, medical insurance, marital status, illness duration, suicide attempt, family history of substance abuse, anxious and atypical features; QIDS-SR16 = 16-item Quick Inventory of Depressive Symptomatology – Self-report Wisniewski et al, The Lancet, in preparation
Adjusted AnalysesAdjusted Analysesbb
OutcomeOutcome OROR (95% CI)(95% CI) PP
QIDS-SRQIDS-SR1616 remissionremission
1.3311.331 (1.073,1.651(1.073,1.651))
0.00930.0093
QIDS-SRQIDS-SR1616 response response 1.3711.371 (1.122,1.675(1.122,1.675))
0.00200.0020
OutcomeOutcome ΒΒ (95% CI)(95% CI) PP
Exit QIDS-SRExit QIDS-SR1616 -0.681-0.681 (-(-1.198,-.165)1.198,-.165)
0.00980.0098
QIDS-SRQIDS-SR1616 % % changechange
-4.276-4.276 (-(-7.424,-.129)7.424,-.129)
0.00780.0078
Phase III clinical trial criteria do not recruit Phase III clinical trial criteria do not recruit samples representative of depressed samples representative of depressed patients who seek treatment in typical patients who seek treatment in typical clinical practice. clinical practice.
The use of broader inclusion criteriaThe use of broader inclusion criteria would make findings more generalizable to would make findings more generalizable to
typical care-seeking outpatientstypical care-seeking outpatients may reduce placebo response and remission may reduce placebo response and remission
rates in Phase III trials, and rates in Phase III trials, and may reduce the risk of failed trials, at the risk of may reduce the risk of failed trials, at the risk of
increasing adverse events and decreasing increasing adverse events and decreasing symptomatic benefit.symptomatic benefit.
What is the pay off?What is the pay off? By any measure, successBy any measure, success
Over 4000 patients involvedOver 4000 patients involved Over 150 cliniciansOver 150 clinicians
Active involvement of PC sitesActive involvement of PC sites 51 publications to date, and more in press or 51 publications to date, and more in press or
preparationpreparation At least 3 large scale ancillary studies (Child, At least 3 large scale ancillary studies (Child,
Alcohol, Genetics), each of which has its own Alcohol, Genetics), each of which has its own cadre of publicationscadre of publications
Depression Treatment Network Depression Treatment Network infrastructure, supporting rapid trial turn infrastructure, supporting rapid trial turn aroundaround
What questions could not be What questions could not be answered?answered?
How does high quality measurement-How does high quality measurement-based care compare to usual care?based care compare to usual care?
Is switching or augmentation the Is switching or augmentation the preferred strategy after 1 or 2 preferred strategy after 1 or 2 failures?failures?
What is the role of cognitive therapy?What is the role of cognitive therapy?
What important questions does What important questions does STAR*D raise?STAR*D raise?
ClinicalClinical Given chronicity and low remission rates of Given chronicity and low remission rates of
most depressions, should combination meds most depressions, should combination meds (“broad spectrum antidepressants”) be started (“broad spectrum antidepressants”) be started at initial treatment step?at initial treatment step?
How do you balance the effort at adequately How do you balance the effort at adequately treating those identified with identifying those treating those identified with identifying those undetected? Could system keep up?undetected? Could system keep up?
Study DesignStudy Design How best do you handle the role of patient How best do you handle the role of patient
preference in study design?preference in study design?
PolicyPolicy Why not include more broadly representative Why not include more broadly representative
patients in placebo-controlled trials used to develop patients in placebo-controlled trials used to develop treatments? treatments?
If you could ensure patient safety and ensure internal If you could ensure patient safety and ensure internal validity in such trials, the results would be more directly validity in such trials, the results would be more directly applicable to our patients, who are less likely to applicable to our patients, who are less likely to spontaneously improve.spontaneously improve.
What should the arsenal of available What should the arsenal of available antidepressants be at the state level?antidepressants be at the state level?
How best do you keep these infrastructures How best do you keep these infrastructures funded?funded?
The STAR*D Study The STAR*D Study InvestigatorsInvestigators National Coordinating CenterNational Coordinating Center
A. John Rush, MDA. John Rush, MD
Madhukar H. Trivedi, MDMadhukar H. Trivedi, MD
Diane Warden, PhD, MBADiane Warden, PhD, MBA
Melanie M. Biggs, PhDMelanie M. Biggs, PhD
Kathy Shores-Wilson, PhDKathy Shores-Wilson, PhD
Diane Stegman, RNCDiane Stegman, RNC
Michael Kashner, PhD, JDMichael Kashner, PhD, JD
Data Coordinating CenterData Coordinating Center
Stephen R. Wisniewski, PhDStephen R. Wisniewski, PhD
G.K. Balasubramani, PhDG.K. Balasubramani, PhD
James F. Luther, MAJames F. Luther, MA
Heather Eng, BA.Heather Eng, BA.
University of AlabamaUniversity of Alabama
Lori Davis, MD Lori Davis, MD
University of California, Los University of California, Los AngelesAngeles
Andrew Leuchter, MDAndrew Leuchter, MD
Ira Lesser, MDIra Lesser, MD
Ian Cook, MDIan Cook, MD
Daniel Castro, MD Daniel Castro, MD
University of California, San University of California, San DiegoDiego
Sidney Zisook, MDSidney Zisook, MD
Ari Albala, MDAri Albala, MD
Timothy Dresselhous, MDTimothy Dresselhous, MD
Steven Shuchter, MDSteven Shuchter, MD
Terry Schwartz, MD Terry Schwartz, MD
Northwestern University Medical Northwestern University Medical School, ChicagoSchool, Chicago
William T. McKinney, MDWilliam T. McKinney, MD
William S. Gilmer, MDWilliam S. Gilmer, MD
The STAR*D Study The STAR*D Study InvestigatorsInvestigators University of Kansas, Wichita University of Kansas, Wichita
and Clinical Research Instituteand Clinical Research Institute
Sheldon H. Preskorn, MDSheldon H. Preskorn, MD
Ahsan Khan, MDAhsan Khan, MD
Massachusetts General Hospital, Massachusetts General Hospital, BostonBoston
Jonathan Alpert, MDJonathan Alpert, MD
Maurizio Fava, MDMaurizio Fava, MD
Andrew A. Nierenberg, MD Andrew A. Nierenberg, MD
University of Michigan, Ann University of Michigan, Ann ArborArbor
Elizabeth Young, MDElizabeth Young, MD
Michael Klinkman, MDMichael Klinkman, MD
Sheila Marcus, MDSheila Marcus, MD
New York State Psychiatric New York State Psychiatric Institute and Columbia Institute and Columbia College of Physicians and College of Physicians and Surgeons, New YorkSurgeons, New York
Frederic M. Quitkin, MDFrederic M. Quitkin, MD
Patrick J. McGrath, MDPatrick J. McGrath, MD
Jonathan W. Stewart, MDJonathan W. Stewart, MD
Harold Sackeim, PhDHarold Sackeim, PhD
University of North Carolina, University of North Carolina, Chapel HillChapel Hill
Robert N. Golden, MDRobert N. Golden, MD
Bradley N. Gaynes, MD Bradley N. Gaynes, MD
The STAR*D Study The STAR*D Study InvestigatorsInvestigators Laureate Healthcare Laureate Healthcare
System, TulsaSystem, TulsaJeffrey Mitchell, MDJeffrey Mitchell, MDWilliam Yates, MDWilliam Yates, MD
University of Pittsburgh University of Pittsburgh Medical Center, Medical Center, PittsburghPittsburghMichael E. Thase, MDMichael E. Thase, MDEdward S. Friedman, MD Edward S. Friedman, MD
Vanderbilt University Vanderbilt University Medical Center, NashvilleMedical Center, NashvilleSteven Hollon, PhDSteven Hollon, PhDRichard Shelton, MDRichard Shelton, MD
The University of Texas The University of Texas Southwestern Medical Southwestern Medical Center, DallasCenter, DallasMustafa M. Husain, MDMustafa M. Husain, MDMichael Downing, MDMichael Downing, MDDiane Stegman, RNCDiane Stegman, RNCLaurie MacLeod, RN Laurie MacLeod, RN
Virginia Commonwealth Virginia Commonwealth University, RichmondUniversity, RichmondSusan G. Kornstein, MDSusan G. Kornstein, MDRobert K. Schneider, MD Robert K. Schneider, MD
Pharmaceutical Industry Pharmaceutical Industry Support for STAR*DSupport for STAR*D
Medications were provided gratis Medications were provided gratis by Bristol-Myers Squibb Company, by Bristol-Myers Squibb Company,
Forest Pharmaceuticals Inc., Forest Pharmaceuticals Inc., GlaxoSmithKline, King GlaxoSmithKline, King
Pharmaceuticals, Organon Inc., Pharmaceuticals, Organon Inc., Pfizer Inc., and Wyeth-Ayerst Pfizer Inc., and Wyeth-Ayerst
Laboratories.Laboratories.
