sheryl l. bishop, phd
DESCRIPTION
Screening for PTSD in a community sample after Hurricane Ike: evaluating the psychometric utility of a brief screening instrument to assess coping, worry, and depression. Sheryl L. Bishop, PhD. Parent Study: The Impact of Stress and Coping on Physical and Mental Health after Hurricane Ike. - PowerPoint PPT PresentationTRANSCRIPT
Screening for PTSD in a community sample after Hurricane Ike: evaluating the psychometric utility of a brief screening instrument to assess coping, worry, and depression
SHERYL L. BISHOP, PHD
2012 National State of the Science Congress on Nursing Research
Parent Study: The Impact of Stress and Coping on Physical and Mental Health after Hurricane Ike
Phase I: At 4-6 months post-hurricane, online baseline assessment of measures of hurricane impact, perceived stress, depression, PTSD, resource loss, resilience, positive growth, physical symptoms, coping, social support, worry, displacement, damage, injury, financial data, health utilization data, employment & work disruption & four biological measures of stress via salivary samples were collected.
SL Bishop, University of Texas Medical BranchRJ Ruiz, University of Texas at Austin
R Stowe, Microgen Labs, IncCR Brewin, University College, London
CC Benight, University of Colorado at Colorado Springs
2012 National State of the Science Congress on Nursing Research
Parent Study: The Impact of Stress and Coping on Physical and Mental Health after Hurricane Ike
Phase II: Repeated assessment of measures of all variables were collected at the 12 month and 24 month anniversary of Hurricane Ike.
2012 National State of the Science Congress on Nursing Research
Specific Aims• Among many foci of interest was the opportunity to validate a
short screening instrument (10 items) for post-traumatic stress symptomology, the Trauma Screening Questionnaire (TSQ) by Brewing et al.1
• Developed from the PSS-SR2 a 17 item instrument• We used a modified PSS-SR, the MPSS-SR3, that included
intensity as well as frequency for each of the 17 items as our ‘gold standard’.
1. Chris R. Brewin, Suzanna Rose, Bernice Andrews, John Green, Philip Tata, Chris McEvedy, Stuart Turner and Edna B. Foa, Brief Screening Instrument for post-traumatic stress disorder, British Journal of Psychiatry (2002), 181, 158-162
2. Foa, E. B., Riggs, D. S., Dancu,C.V., et al (1993) Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. Journal of Traumatic Stress, 6, 459-473.
3. Falsetti, S. A., Resnick, H. S., Resick, P. A., & Kilpatrick, D. (1993). The Modified PTSD Symptom Scale: A brief self-report measure of posttraumatic stress disorder. The Behavioral Therapist, 16, 161-162.
2012 National State of the Science Congress on Nursing Research
Sample
• UTMB community: Faculty, staff and students– 7th largest employer in the region (13,000 employees
statewide) – 6 hospitals– State’s oldest schools of Medicine, Nursing, Health
Professions and Biomedical Sciences– 2,338 students(2,094 from 122 TX counties), over 700
interns, residents, and fellows, and 1,116 faculty.• 391 respondents with complete data for this analysis
Male15%
Female 85%
Gender
White77%
Black or African Ameri-
can1%
Hispanic12%
Asian8%
American Indian or Alaska Native
1%
Males by Race
White83%
Black or African Ameri-
can3%
Hispanic10%
Asian3%
American Indian or Alaska Native
1%
Females by Race
N=477 Respondents
Displac
ed
Damag
eInjury
Inconve
nience
Family
Community1
1.5
2
2.5
3
3.5
4
4.5
5
Hurricane Exposure Ratings by Gender
MalesFemales
A great deal
None
2012 National State of the Science Congress on Nursing Research
Instrument CharacteristicsMean Sd Range Items Scale
TSQ 3.35 2.65 0-10 10 Yes (at least twice a week)/no
MPSS-SR 20.06 17.46 0-104 17 0 (not at all)- 3 (5 or more times a week)
Criterion for affirmative PTSD symptomology for TSQ is a score of 6 and 46 for the MPSS-SR.
2012 National State of the Science Congress on Nursing Research
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 52 57 65 72 940
5
10
15
20
25
30
MPSS-SR
Fre
qu
ency
8.2%
2012 National State of the Science Congress on Nursing Research
21.2%
N=401
2012 National State of the Science Congress on Nursing Research
N=393
48.8%
34.6%
N=393
2012 National State of the Science Congress on Nursing Research
Validating
• Correlation TSQ x MPSS-SR=.767, p<.001However correlation only indicates that a positive relationship exists. It cannot indicate whether evaluations AGREE. Therefore, it is NOT the appropriate approach.
2012 National State of the Science Congress on Nursing Research
Indicators of Agreement• Interclass correlation coefficient – ICC
– Must transform both instruments into the same scale metric…z scores computed for this purpose.
– Indicates moderately high agreement however, Bland and Altman (1990) provide compelling arguments against the use of the ICC to evaluate whether two methods are equivalent.
Intraclass Correlation Coefficient: Two-way random effects modelICC 95% Confidence Interval F Test
Lower Bound Upper Bound Value SigSingle Measures .767 .723 .805 7.581 .000Average Measures .868 .839 .892 7.581 .000
2012 National State of the Science Congress on Nursing Research
Crosstabulation Analyses• Kappa – compares rating agreement when
variables are dichotomized into DX = Yes/No.– Clearly the criteria used to determine cut-offs is
critical– And when evaluating a new instrument against a
parent instrument, the assumption that the parent instrument is ‘true’ may be unsupported. Thus conclusions about concurrence may be biased in favor of the reference instrument.
2012 National State of the Science Congress on Nursing Research
•Kappa =.414, p<.001 is moderate agreement•Sensitivity (True Positives)=78.95%•Specificity (True Negatives)=84.7%•False Positives (Type I error)=15.3%•False Negatives (Type II error)= 21.1% •PV+=35.7%•PV-=97.4%
T1ptsd (reference) Total
DX+ (yes) DX- (no) t1tsq DX+(yes) 30 54 84 DX-(no) 8 299 307
Total 38 353 391
2012 National State of the Science Congress on Nursing Research
Distribution of Difference Scores
d=-.0102sd=.68153
Limits of Agreement:d-2s = -.0102-(1.36306)= -1.37326d+2s=-.0102+(1.36306)= 1.35286N=13 N=7
2012 National State of the Science Congress on Nursing Research
Convergent Validity
• Forward stepwise regressions were conducted with the Depression the as outcome variable
• The predictor set was hurricane exposure scores, worry, 14 coping styles and either the TSQ or the MPSS-SR PTSD scores
• Convergent validity would be demonstrated if both instruments comparably predicted depression.
2012 National State of the Science Congress on Nursing Research
Predicting DepressionTSQ: R2=.672 MPSS-SR: R2=.684
PREDICTORS t Partial t Partial
PTSD SCORE 8.484*** .405 14.322*** .598
Self Blame 5.774*** .289 3.855*** .197
Worry 5.799*** .290 5.425*** .272
Behavioral Disengage.
4.345*** .221 2.063* .107
Positive Reframing
-3.454*** -.177 -4.669*** -.236
Hurricane Exposure
2.688** .139 ns --
Humor -2.284* .118 ns --
2012 National State of the Science Congress on Nursing Research
Summary
• All ‘traditional’ assessments of equivalency show a high degree of ‘agreement’ over response patterns or classifications.
• Newer procedures also support a moderate to high degree of equivalency between the two instruments.
• Limits of agreement indicate a fairly narrow range of variability between assessments between the two instruments.
2012 National State of the Science Congress on Nursing Research
Summary Continued• TSQ does a fairly good job of identifying both true
positives and true negatives. • False positive and false negative rates indicate that
TSQ missed more people WITH PTSD than incorrectly diagnosing those as having PTSD who did not suffer
• High Negative Predictive value = 97.4% = accurate in identifying those that are NOT likely to be suffering from PTSD,
• Low Moderate Positive Predictive value = 35.7% = fair at identifying those that definitely have symptoms of PTSD.
2012 National State of the Science Congress on Nursing Research
Summary continued
• Regression analyses further demonstrated convergent validity by highly similar and reflective patterns of prediction of depression.
• There is some indication of support for more complex modeling with the use of the TSQ to allow for finer distinctions in coping behaviors.
2012 National State of the Science Congress on Nursing Research
Conclusion
• The TSQ appears to be a good screening tool for use in early triage situations where false positives are more preferable to false negatives. It has high negative predictive value and can reliably rule out those individuals that do not exhibit PTSD symptoms but may incorrectly classify some as having symptoms that will later be ruled out on further testing.
QUESTIONS?