A Cross-sectional, Randomized, Non-interventional Methods Study to Compare Three Methods of Assessing Suicidality in Psychiatric Inpatients

Similar documents
Suicidal Ideation & Behavior Discussion. Roger E. Meyer, MD Professor of Psychiatry Penn State Hershey Medical Center

Methodological approach to the use and interpretation of baseline history of suicidal ideation and behavior Sarah DuBrava Pfizer, Inc

Clinical Trial Database Analyses to Inform Regulatory Guidances Suicidal Ideation and Behavior Assessment

Examining Inter-Rater Reliability of a CMH Needs Assessment measure in Ontario

(true) Disease Condition Test + Total + a. a + b True Positive False Positive c. c + d False Negative True Negative Total a + c b + d a + b + c + d

4 Diagnostic Tests and Measures of Agreement

Repeatability of a questionnaire to assess respiratory

Screening for Depression and Suicide Risk Assessment

ISCTM Suicidal Ideation & Behavior Assessment Working Group

PROVIDED IN RESPONSE TO YOUR UNSOLICITED REQUEST FOR INFORMATION

Linking Assessments: Concept and History

Supplementary Material

COMMITMENT &SOLUTIONS UNPARALLELED. Assessing Human Visual Inspection for Acceptance Testing: An Attribute Agreement Analysis Case Study

COMPUTING READER AGREEMENT FOR THE GRE

Figure 1: Design and outcomes of an independent blind study with gold/reference standard comparison. Adapted from DCEB (1981b)

ASD Working Group Endpoints

May and Klonsky s (2016) meta-analysis of factors

Appendix table 1. Gender differences in lifetime prevalence, inter-quartile range. and ageof-onset of DSM-IV/CIDI bipolar disorder (BPD)

Posttraumatic Stress Disorder and Suicidal Behavior: Current Understanding and Future Directions

Suicidal and Self-Injurious Behaviors in Youth with Borderline Personality Disorder (BPD) Features

Reliability. Internal Reliability

7/17/2013. Evaluation of Diagnostic Tests July 22, 2013 Introduction to Clinical Research: A Two week Intensive Course

2 Philomeen Weijenborg, Moniek ter Kuile and Frank Willem Jansen.

National Academy of Science July 17-18, 2018 Washington DC Larry Alphs, MD, PhD RESTRICTION OF TREATMENT QUALITY IN PRAGMATIC CLINICAL TRIALS

Identifying Adult Mental Disorders with Existing Data Sources

1. Introduction. 2. Objectives. 2.1 Primary objective

BUILDING BARRIERS TO SUICIDE:

Introduction to ROC analysis

Sensitivity and specificity of depression screening tools among adults with intellectual and developmental disabilities (I/DD)

Brief Report: Interrater Reliability of Clinical Diagnosis and DSM-IV Criteria for Autistic Disorder: Results of the DSM-IV Autism Field Trial

Kim L. Gratz Department of Psychiatry and Human Behavior University of Mississippi Medical Center (UMMC)

Week 2 Video 2. Diagnostic Metrics, Part 1

reproducibility of the interpretation of hysterosalpingography pathology

Evaluating Quality in Creative Systems. Graeme Ritchie University of Aberdeen

U.S. 1 February 22, 2013

How do I do a proper suicide assessment and document it in my note? September 27, 2018

Tubal subfertility and ectopic pregnancy. Evaluating the effectiveness of diagnostic tests Mol, B.W.J.

Importance of Good Measurement

When first published in 1992, the Research Diagnostic

NIH Public Access Author Manuscript J Clin Psychiatry. Author manuscript; available in PMC 2014 April 07.

Juvenile Pre-Disposition Evaluation: Reliability and Validity

Substance Abuse Questionnaire Standardization Study

Guidelines for using the Voils two-part measure of medication nonadherence

DESCRIPTION OF FOLLOW-UP SAMPLE AT INTAKE SECTION TWO

[ O R I G I N A L R E S E A R C H ] Innov Clin Neurosci. 2014;11(9 10):23 31

Pain Assessment in Elderly Patients with Severe Dementia

Small-area estimation of mental illness prevalence for schools

Forecasting Depression: Past, Present, and Future Mental Health in Los Angeles County

A comparison of diagnosis of dementia using GMS AGECAT algorithm and DSM-III-R criteria

Suicidal Ideation and Behavior Working Group:

Santa Fe Municipal Court P.O. Box 909 Santa Fe, New Mexico 87504

Relationship Between Intraclass Correlation and Percent Rater Agreement

Comparison of the 32-item Hypomania Checklist, the 33-item Hypomania Checklist, and the Mood Disorders Questionnaire for bipolar disorder

Using the NWD Integrative Screener as a Data Collection Tool Agency-Level Aggregate Workbook

State of Iowa Outcomes Monitoring System

IMPLEMENTING STANDARDS FOR QUALITY PALLIATIVE CARE. Monica Malec, MD University of Chicago

Predictors of repeated self-harm in older New Zealanders

Ofir Turel. Alexander Serenko

A STUDY INTO THE THEMES AND TYPOLOGIES OF BINGE GAMBLING EPISODES BY VERITY HARRIS. SUPERVISOR: DR. AMANDA ROBERTS AND DR STEPHEN SHARMAN

With additional support from Florida International University and The Children s Trust.

Study No.: Title: Rationale: Phase: Study Period: Study Design: Centers: Indication: Treatment: Objectives: Primary Outcome/Efficacy Variable:

Suicide Prevention in the United States: Challenges, Opportunities, and Innovations. Richard McKeon Ph.D.Chief, Suicide Prevention Branch SAMHSA

Small-area estimation of prevalence of serious emotional disturbance (SED) in schools. Alan Zaslavsky Harvard Medical School

BEHAVIORAL ASSESSMENT OF PAIN MEDICAL STABILITY QUICK SCREEN. Test Manual

Gambler Addiction Index: Gambler Assessment

? Health Care System Research Network Conference April 9, 2019 Julie Richards, PhDc, MPH

Study Endpoint Considerations: Final PRO Guidance and Beyond

Parental Perception of Quality of Hospital Care for Children with Sickle Cell Disease

Research with the SAPROF

Hopelessness Predicts Suicide Ideation But Not Attempts: A 10-Year Longitudinal Study

Chapter Two. Classification and treatment plans

Victim Index Reliability and Validity Study

ISCTM SIB Consensus Statements for the Nomenclature & Classification Working Group

Individuals with Psychiatric and Substance Use Disorders Comprise an Important Segment of Smokers & Consume Nearly 1 in 2 Cigarettes Sold

ALABAMA SELF-ASSESSMENT INDEX PILOT PROGRAM SUMMARY REPORT

English 10 Writing Assessment Results and Analysis

10/7/2016. Janet Kittams-Lalley Helpline Center

MAPS Study MP-10 1 Study Synopsis UK April 18, 2011

No part of this page may be reproduced without written permission from the publisher. (

5/11/2017. Objectives. Epidemiology of suicidality in youth. Suicide in Children & Adolescents: Risks, Screening & Prevention

Issues in Clinical Measurement

Validating a computable phenotype: Should results change a trial s primary outcome?

Columbia-Suicide Severity Rating Scale Baseline (C-SSRS BASELINE)

The Difficult Patient. Psychiatric Dilemmas in the Primary Care Setting. No Disclosures. Objectives 10/12/17. Erick K. Hung, MD

Suicide Risk Assessment

Spirometry is the most frequently performed. Obstructive and restrictive spirometric patterns: fixed cut-offs for FEV1/FEV6 and FEV6

Centerstone Research Institute

non suicidal self injury does social support make a difference an epidemiological investigation of a danish national sample

Psychotherapy research historically focused predominantly

Update on the Reliability of Diagnosis in Older Psychiatric Outpatients Using the Structured Clinical Interview for DSM IIIR

A Comparison of Diagnostic Interviews for Children. A Senior Honors Thesis

Patient characteristics. Intervention Comparison Length of followup. Outcome measures. Number of patients. Evidence level.

Statistical Validation of the Grand Rapids Arch Collapse Classification

MINDFUL WELLNESS CENTER, PLLC

QUALITY ASSESSMENT (QA) MANUAL

Chapter 3. Psychometric Properties

Clinical and public health significance of the proposed changes to the diagnosis of alcohol abuse and dependence in DSM-5

Review of Various Instruments Used with an Adolescent Population. Michael J. Lambert

Validity and reliability of measurements

Screening (Diagnostic Tests) Shaker Salarilak

Transcription:

A Cross-sectional, Randomized, Non-interventional Methods Study to Compare Three Methods of Assessing Suicidality in Psychiatric Inpatients Eric A. Youngstrom, Ph.D., Ahmad Hameed, M.D., Michael Mitchell, B.A., Andrew J. Freeman, M.A., Anna Van Meter, M.A., Guillermo Perez Algorta, Ph.D., Alan Gelenberg, M.D., & Roger E. Meyer, M.D.

Acknowledgments Study supported by an investigator initiated award from Pfizer, Inc. to Penn State Hershey Medical Center Alan Gelenberg, MD PI

Primary objective: Compare self-report and clinical interviewbased versions of the older and current editions of the Sheehan Suicide Tracking Scale (S-STS) to the C-SSRS in the assessment of suicidal ideation/behavior in the context of the domains defined by the C-CASA and in relationship to the clinical evaluation of recently admitted psychiatric inpatients.

"'In this world nothing can be said to be certain, except death and taxes." -Benjamin Franklin, 1789 I am realizing Franklin was not a clinical trialist.

C-CASA Classification (circa 2010) Item Classification Mapping 1 Completed suicide Impossible to interview 2 Suicide Attempt CCASA #2 3 Preparatory suicidal behavior CCASA #3 4 Suicidal Ideation CCASA #4 5 Other, no deliberate self harm NA in CSSRS interview 6 Fatal, but not enough information to classify Impossible to interview 7 Self-injurious act, no suicidal intent (NSSI) CCASA #7 8 Self-injurious behavior, intention unknown Interview should not have unknown 9 Nonfatal, not enough information to classify

C-CASA Classification (August 2012)

Secondary Objectives Quantify relative ease, time and means of administration (self-report and clinical interview) Inter-rater and Intra-rater reliability (over time) of the scoring of rater administered form Patient preference re: acceptability of the instruments

Urn Randomization Stratified variables: Age (up to 24 years versus 25 and older) Psychosis (present/absent) Two arms: C-SSRS, new S-STS self report, old S-STS interview C-SSRS, old S-STS self report, new S-STS interview Order counterbalanced within each arm 6 sequences

Other design features Break between interviews Same day completion whenever possible Same interviewer Video recording for re-rating by different judge (n = 90 will be re-rated) Raters trained by Posner and by Sheehan

Demographics (N =199) Variable Descriptive Age M = 38.5 (SD = 12.4) 18 to 65 years Gender 113 female (57%) Race 157 White (79%), 37 Black (18%), 9 American Indian (5%), 6 Asian/Pacific (3%) Hispanic 26 (13%) Married 40 (20%) Education 87% with GED or higher Raters 3 interviewers, 3 different video raters Videos 60 rated (more than 90 available)

Administration Times C-SSRS > New S-STS > Old STS all p <.001 Five minutes or less (ecssrs sig faster)

Minutes to Administer C-SSRS Old STS New S-STS Minimum 3 3 3 25 th Pct 12 4.3 11 Median 16 6 15 75 th Pct 22 7 20 Maximum 57 20 40 Mean (SD) 17.8 (8.7) 6.5 (2.7) 16.2 (7.4) N = 199

Interview Length Much Longer if Endorse Suicidal Ideation

What is a good kappa? K 1.0.9.8.7.6.5.4.3.2.1.0 Excellent Good Fair Poor Cicchetti & Sparrow, 1981 Excellent Fair to Good Poor Fleiss, 1981 Almost Perfect Substantial Moderate Fair Slight Poor Landis & Koch (1977) (Excellent) Very Good Good Questionable Unacceptable Regier et al. (2012) DSM-5

Reliability Standoff Two sources of error affect each estimate: Reliability of Test A, Reliability of Test B C-SSRS Kappa Kappa New S-STS Kappa Old STS

Kappa for Lifetime Attempts (C-CASA #2).68 (.06) C-SSRS.66 (.06) New S-STS.68 (.06) Old STS

Recommended approach for missing or imperfect gold standard Pepe, M. S. (2003). The statistical evaluation of medical tests for classification and prediction. New York: Wiley. Zhou, X.-H., Obuchowski, N. A., & McClish, D. K. (2002). Statistical methods in diagnostic medicine. New York: Wiley.

LCA Approach Put all converging measures into analysis Fit two class solution: Diagnosis+ vs. Diagnosis-, Haves vs. Have Nots Save predicted class membership Can estimate sensitivity and specificity of each measure calibrated against latent class (could also test larger number of classes, but diagnostic efficiency statistics no longer meaningful)

Latent Class Model True status, cannot measure perfectly Latent Class Drives performance on measured variables C-SSRS New S-STS Old STS

Latent Class Model Suicidal Behavior (C-CASA 2) Lifetime Kappa Sensitivity Specificity C-SSRS.82.93.94 New S-STS.86.91.85 Old STS.83.97.85 C-CASA 2 Behavior None perfect, all excellent C-SSRS New S-STS Old STS

Latent Class Model Preparatory Acts (C-CASA 3) Lifetime C-CASA 3 Prep acts Kappa C-SSRS.86 New S-STS.73 Old STS.63 Weakest agreement across the measures C-SSRS New S-STS Old STS

Latent Class Model Ideation (C-CASA 4) Lifetime Kappa Sensitivity Specificity C-SSRS.72.99.80 New S-STS.91.99 1.00 Old STS.83.99 1.00 C-CASA 4 Ideation C-SSRS New S-STS Old STS

Latent Class Model NSSI (C-CASA 7) Lifetime C-CASA 7 NSSI Kappa C-SSRS.78 New S-STS.77 Old STS.83 C-SSRS New S-STS Old STS

Kappas for Lifetime Ratings 1.0.9.8.7 K.6.5 New S-STS vs. C-SSRS Old S-STS vs. C-SSRS.4 New S-STS vs. Old S-STS.3.2.1.0 CASA-2 Attempts CASA-3 Prep CASA-4 Ideation CASA-7 NSSI

Kappas for Lifetime Ratings 1.0.9.8 K.7.6.5.4.3 New S-STS vs. C-SSRS Old S-STS vs. C-SSRS New S-STS vs. Old S-STS New S-STS vs. LCA CSSRS vs. LCA Old S-STS vs. LCA.2.1.0 CASA-2 Attempts CASA-3 Prep CASA-4 Ideation CASA-7 NSSI

Kappas for Lifetime Ratings 1.0.9.8 Excellent K.7.6.5 Good Fair.4.3 Poor.2.1.0 Attempts CASA-2 Prep CASA-3 Ideation CASA-4 NSSI CASA-7 New S-STS vs. LCA CSSRS vs. LCA Old S-STS vs. LCA

Kappas for Lifetime Ratings across 4 C-CASA categories 1.0.9.8 K.7.6.5.4.3 New S-STS vs. C-SSRS Old S-STS vs. C-SSRS New S-STS vs. Old S-STS New S-STS vs. LCA CSSRS vs. LCA Old S-STS vs. LCA.2.1.0 Attempts CASA-2 Prep CASA-3 Ideation CASA-4 NSSI CASA-7

Summary Suicidal Behavior Past Month N=73 (37%) Kappa BCa Low High Sensitivity Specificity NPV PPV CSSRS.901.835.958.904.984.947.971 New S-STS.915.852.966 1.000.937 1.000.901 Old S-STS.640.544.724.945.746.959.683 Lifetime N=152 (76%) CSSRS.818.720.908.928.938.804.979 New S-STS.859.772.940.980.854.932.955 Old S-STS.833.726.918.967.854.891.955 Median.85.75.93.96.90.94.96 All values sig p <.0005

Summary Suicidal Ideation Past Month N=180(90%) Kappa BCa Low High Sensitivity Specificity NPV PPV CSSRS.784.637.898.950 1.000.679 1.000 New S-STS.778.615.902.972.842.762.983 Old S-STS.843.672.965.789.994.938.978 Lifetime N=195(98%) CSSRS.720.219 1.000.990.800.667.995 New S-STS.907.660 1.000.995 1.000.833 1.000 Old S-STS.828.492 1.000.990 1.000.714 1.000 Median.81.63.98.98 1.00.74 1.00 All values sig p <.0005

Behavior: Kappa and 95% c.i. (BCa) K 1.0.9.8.7.6.5.4.3.2.1.0 Past Month Lifetime

Ideation: Kappa and 95% c.i. (BCa) K 1.0.9.8.7.6.5.4.3.2.1.0 Past Month Lifetime

Inter-Rater Reliability (more tape coding planned a priori) Lifetime C-SSRS (N =60) New S-STS (n =27) Old STS (n =33) CASA #2 (Attempt).893 (.060) <.0005.658 (.181).001.857 (.098) <.0005 CASA #3 (Prep).651 (.106) <.0005.257 (.178) ns.857 (.097) <.0005 CASA #4 (Ideation).551 (.226) <.0005 -- a -- a CASA #7 (NSSI).799 (.078) <.0005.703 (.137) <.0005.767 (.106) <.0005 Past Month CASA #2 (Attempt).720 (.097) <.0005.684 (.144) <.0005.937 (.062) <.0005 CASA #3 (Prep).800 (.077) <.0005.488 (.163).009.879 (.082) <.0005 CASA #4 (Ideation) 1.000 (.000) <.0005.867 (.129) <.0005 1.000 (.000) <.0005 CASA #7 (NSSI).795 (.114) <.0005.807 (.131) <.0005.841 (.155) <.0005 a Cannot estimate -- 100% ideation on cases rated

Preliminary Inter-Rater Reliability CASA 2012 Categories C-SSRS (N =60) CASA 2012 Category C-SSRS Lifetime C-SSRS Past Month #1 Passive Ideation.734 (.178).924 (.075) #2 Ideation (no method, intent, plan).703 (.160).957 (.043) #3 Ideation+Method (no intent, plan).774 (.124).926 (.052) #4 Ideation+Method+Intent (no plan).884 (.065).934 (.046) #5 Ideation+Method+Intent +Plan.770 (.081).688 (.093) #7 Attempt.893 (.060).720 (.097) #8 Interrupted Attempt.678 (.103).573 (.137) #9 Aborted Attempt.602 (.101).946 (.053) #10 Prep Acts.761 (.085).804 (.094) #11 NSSI.799 (.078).795 (.114) No completed suicides (CASA 2012 #6); all K significant p <.0005

Patient Acceptance: Majority highly satisfied 100 90 80 70 60 50 40 30 20 10 0 d =.08 d =.40 Patient Acceptance (% Max) CSSRS New S-STS CSSRS Old STS

Key Points Validity of all three versions excellent Using better model yields more accurate estimate Agreement with each other good Inter-rater reliability (video) excellent Similar patient, interviewer preferences & tolerability Multiple endpoints complicates analyses and interpretation

Limitations Did not compare CSSRS interview to E version Inpatient unit meant extremely high rate of ideation Important clinical setting & context Challenging for tests (big standard errors) Consumer Reports model generalization across clinical settings vital Need to explore moderators of accuracy Did not assess prospective prediction or treatment response

Future Directions Investigate where the signal is Brown s talk severe ideation more important than mild Start with simple models and measures Have data guide decisions to add complexity

Complexity has a cost Categories df Type I Error Rate Evidence about Model Suicidal? Yes/No 1.05 Probably oversimplified Ideation, Behavior 2.10 (Could differentiate severe ideation from mild ) C-CASA 2010, Interview 4.19 C-CASA 2012 11.43 May be too complex Broadly read C-CASA 2012 15.54 Definitely too complex Endpoints probably not all equally important clinically Cannot easily adjust Type II error too serious What are costs of false positive in drug development or post marketing?

"Everything should be made as simple as possible, but not simpler." -Albert Einstein, 1933

Thank You!

Kappa for Lifetime Preparatory Acts (C-CASA #3).59 (.06) C-SSRS.47 (.07) New S-STS.39 (.06) Old STS

Kappa for Lifetime Ideation (C-CASA #4).66 (.19) C-SSRS.60 (.19) New S-STS.76 (.13) Old STS

Kappa for Lifetime NSSI (C-CASA #7).56 (.06) C-SSRS.61 (.05) New S-STS.61 (.06) Old STS