existing sources of epidemiological data on...

33
Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D. and John Boyle Ph.D. Medical U. of South Carolina ICF International Workshop on Integrating New Measures of Trauma into SAMSHA’s Data Collection Systems, 12/17/2015

Upload: others

Post on 16-Mar-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Existing Sources of

Epidemiological Data on Trauma

Dean G. Kilpatrick Ph.D. and John Boyle Ph.D.

Medical U. of South Carolina ICF International

Workshop on Integrating New Measures of Trauma

into SAMSHA’s Data Collection Systems, 12/17/2015

Page 2: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Overview of Presentation

Key definitional and methodological issues important for

understanding epidemiological research on exposure to

potentially traumatic events (PTEs) and PTSD.

Review of three major national epidemiological studies that

produced prevalence estimates of PTE exposure and PTSD

among probability samples of adults including the NCS-R, the

National Survey on Alcohol and Related Disorders, and the 2008-

2012 National Survey on Drug Use and Health (NSDUH) Mental

Health Surveillance Study (MHSS).

Limitations of existing data from these studies to produce up-to-

date national and state level PTE and PTSD prevalence estimates

using DSM-5 criteria.

Page 3: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Overview (Continued)

Description of recently completed national study of PTE

and PTSD prevalence that demonstrates the feasibility of

collecting PTE and PTSD prevalence data using web-

based data collection.

Suggestions for how to integrate new measures of PTE

exposure and PTSD prevalence into SAMHSA’s data

collection programs and generate national and state-level

PTE and PTSD prevalence estimates using DSM-5 criteria.

Page 4: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

What Does Trauma Mean?

Trauma has been used in two ways, sometimes as a STIMULUS

(e.g. a stressor event thought to be capable of having negative

effects on mental health and behavior)and sometimes as a

RESPONSE (e.g. responses of PTSD or related disorders that

occur following exposure to these stressor events).

When we say “new measures of trauma”, do we mean new

measures of exposure to stressor events, new measures of

responses following exposure to those stressor events, or both?

Page 5: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Importance of Stressor Events in the PTSD Diagnosis

PTSD Criterion A defines types of stressor events capable of

producing PTSD. If a stressor is not a Criterion A event, it

cannot, by definition, produce PTSD, so other PTSD criteria are

not assessed.

Many researchers call Criterion A events Traumatic Events

(TEs) or Potentially Traumatic Events (PTEs). PTE is a better

term because not everyone exposed develops PTSD, so events

are only potentially traumatic.

PTSD Criterion A definitions of PTEs differ in the DSM-III,

DSM-III-R, DSM-IV, and DSM-5, making it difficult to

compare PTE exposure or PTSD prevalence in studies

measuring PTEs or PTSD symptoms using different DSMs.

Page 6: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Key Methodological Issues for Measuring PTE and PTSD Prevalence

A challenge for epidemiological studies is to collect data in the

most cost effective way using methods that facilitate willingness

to disclose information about exposure to all relevant PTEs

including those involving sensitive topics.

A critical issue is whether the survey measures all relevant PTEs

using sensitive behaviorally specific questions to assess PTEs

with the highest probabilities of increasing risk of PTSD (e.g.

PTEs involving sexual violence, other interpersonal violence,

and military combat). If not, these PTEs are undetected, and

PTSD prevalence cannot be assessed properly.

Another challenge is measuring PTSD using current DSM PTSD

criteria that are capable of producing adequate estimates of

Partial, Subthreshold, or Subclinical PTSD.

Page 7: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Measuring Sexual Violence and Other Interpersonal Violence PTE Exposure

PTEs involving sexual violence, intimate partner violence, and

other interpersonal violence are often defined stereotypically by

respondents; they are extremely prevalent and stigmatizing, and

they are more difficult to measure than other PTEs.

Well-designed studies (e.g. CDC National Intimate Partner and

Sexual Violence Survey) indicate that 18% of adult women and

1.4% of adult men in the US have been victims of rape and that

35.6% of women and 28.5% of men have been victims of rape,

physical violence, or stalking by an intimate partner.

Research documents that you can’t measure these PTEs

properly with simple gatekeeping Qs; you need behaviorally

specific Qs that get around stereotypes.

Page 8: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Four Steps in Responding to Survey Questions

Comprehending the question and instructions

Retrieving specific memories or information relevant to the

question

Making judgments about whether the specific memories or

information match what is being asked in the question

Formulating a response based on whether the response is

accurate and other factors including concerns about stigma or

confidentiality

Source: Tourangeau, 1984; NRC Report on Estimating the

Incidence of Rape and Sexual Assault, 2014.

Page 9: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Three Major Studies Producing National PTE and PTSD Estimates

National Comorbidity Survey-Replication (NCS-R)

National Epidemiological Survey on Alcohol and Related

Conditions (NESARC)

National Survey on Drug Use and Health (NSDUH)

Mental Health Surveillance Study (MHSS)

Page 10: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

NCS-Replication (NCS-R):

Conducted in early 2000s as a follow-up to original NCS

which was conducted in early 1990s.

National probability sample of English-speaking adults age

18 and older (N=5692) in the coterminous U.S.

Data on DSM-IV diagnoses were collected in face-to-face

interviews using the WHO-CIDI, a fully-structured interview

conducted by lay interviewers.

Lifetime PTE exposure was assessed comprehensively with a

series of 26 Qs about exposure to specific DSM-IV Criterion

A1 PTEs as well as follow-up Qs to find out which A1 PTE’s

also met A2 by resulting in the respondent having been

terrified or frightened, helpless, shocked or horrified, or

numb.

Page 11: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

NCS-R PTE and PTSD Prevalence Findings

79.4% of the NCS-R sample had been exposed to one or

more DSM-III-defined PTE.

Lifetime PTSD prevalence using DSM-IV criteria was 6.8%

overall; 9.7% among women and 3.6% among men.

Past 12 months PTSD prevalence was 3.5% overall; 5.2%

among women and 1.8% among men.

Page 12: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

National Epidemiological Survey on Alcohol and Related Conditions

NESARC study conducted in 2004-05 measured Full and

Partial PTSD in a nationally representative household

probability sample of 34,653 U.S. adults using the AUDADIS-

IV fully-structured interview.

Conducted by lay interviewers in person.

Assessed lifetime PTE exposure and PTSD using DSM-IV

criteria.

PTE exposure assessed with 27 Qs enumerating specific

PTEs falling under the DSM-IV definition. If they had more

than one PTSD, they identified the one that was the worst

The PTSD module measured all PTSD symptoms with no

skip outs and also measured functional impairment.

Page 13: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

NESARC PTE and PTSD Prevalence Findings

79.7% of respondents were exposed to at least one PTE.

Lifetime prevalence of PTSD to the only or worst PTE was

6.4%.

Partial PTSD was defined as not meeting full diagnostic

criteria for PTSD but having at least one symptom in each

of the B, C, and D Criteria. Lifetime prevalence of Partial

PTSD was 6.6%.

Respondents with both Full PTSD and Partial PTSD had

elevated rates of lifetime mood, anxiety, substance use

disorders, and suicide attempts.

Page 14: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

2008-12 NSDUH Mental Health Surveillance Study: Methodology

Subset of English-speaking adults 18 and older who

completed the NSDUH in person survey were recruited for

the MHSS (N=5652), which assessed selected mental

disorders using DSM-IV criteria.

Mental health professionals conducted survey via telephone

and assessed for mental disorders using modified version

of the SCID-I, a semi-structured interview.

Page 15: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

MHSS PTE and PTSD Assessment

PTE exposure and PTSD assessed using DSM-IV criteria.

Lifetime exposure to PTEs measured using three stage

process. In Stage 1, respondents were asked two

gatekeeping screening questions about exposure to DSM-

IV Criterion A1 events. Stage 2 asked if they ever had re-

experiencing symptoms to any events. Stage 3 determine if

events met the A2 requirement that the event produced

“fear, helplessness, or horror”.

PTSD was assessed to the PTE they considered to have

affected them the most.

Page 16: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

MHSS PTE Screening Questions

“Sometimes things happen to people that are extremely

upsetting- things like being in a life-threatening situation like a

disaster, very serious accident or fire; being physically assaulted

or raped; seeing another person killed or dead, or badly hurt; or

hearing about something horrible that happened to someone you

are close to. At anytime during your life, have any of these things

happened to you”?

If no:

“Have you ever been in any serious car accidents or have you

been victim of a crime”?

If no to both: skip out of PTSD module

Page 17: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

MHSS PTSD Symptom Assessment

PTSD symptom assessment module had numerous skip outs:

If less than 1 Criterion B re-experiencing symptom, skip

out.

If less than 3 Criterion C avoidance/numbing systems, skip

out.

If less than 2 Criterion D hyperarousal symptoms, skip out.

Page 18: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

MHSS Definition of Clinical PTSD and Subclinical PTSD

Clinical PTSD: cases that met all DSM-IV diagnostic criteria

for PTSD.

Definition of Subclinical PTSD includes: 1) cases that did not

meet criteria for Clinical PTSD but that had at least 1 Criterion

C (i.e. Avoidance/Numbing )symptom; AND 2) Clinical PTSD

cases.

Note: the definition of Subclinical PTSD in the MHSS is

confusing because it includes Clinical PTSD.

Page 19: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Summary of DSM-IV PTSD Findings from These Studies

The NCS, NCS-R and NESARC studies had more thorough and

comprehensive assessment of PTE exposure than the MHSS

and found approximately twice the lifetime PTE exposure

prevalence (i.e. 79.4% in the NCS and 79.7% in the NESARC ).

Past 12 month PTSD prevalence was approximately 5 times

lower in the MHSS than in the NCS-R , likely resulting from the

MHSS’s poor screening for PTEs, not administering the PTSD

module to many respondents who likely had undetected PTEs,

and possibly due to error variance introduced by clinician

administered semi-structured interviews.

Skip outs in the MHSS PTSD assessment made its estimate of

Subclinical PTSD severely flawed; the NESARC Partial PTSD

measure was better.

Page 20: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

What about PTSD in DSM-5? Is It Really Different?

There have been changes in Criterion A, so DSM-IV

estimates of the Prevalence of PTE exposure are no longer

applicable.

There have also been other changes in the other PTSD

symptoms and diagnostic criteria, so DSM-IV estimates of

PTSD prevalence are no longer applicable.

Page 21: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Major Changes in DSM-5 Criterion A PTEs

PTEs no longer have to produce “fear, helplessness, or horror

(the A2 Criterion of DSM-IV), a Godsend for epi research!

Types of sexual violence events defined as PTEs were

expanded.

Learning about nonviolent, sudden, unexpected deaths of close

family members or friends is no longer a PTE unless the death

was violent or accidental.

New category of PTEs added involving work-related repeated or

extreme indirect exposure to aversive details of PTEs

experienced by others.

Explicit recognition that exposure to multiple PTEs is common

and that PTSD can occur in response to more than one PTE.

Page 22: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Other DSM-5 PTSD Changes

Three new symptoms (D3, D4, and E2) added and

four others (D1, D2, D7, and E1) modified.

Symptom-based criteria restructured from three in

DSM-IV to four in DSM-5.

Nonspecific PTSD symptoms required to develop or

worsen after exposure to PTE(s).

Acknowledgement that PTSD symptoms can

incorporate responses to more than one traumatic

event.

Page 23: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Obtaining DSM-5 PTSD Prevalence

Data: A Feasibility Online Survey

In conjunction with the DSM-5 PTSD Workgroup, we

developed a web-based assessment instrument designed to

collect data to evaluate the impact of proposed diagnostic

changes on PTSD prevalence.

This web-survey assessment tool was used with two samples.

Page 24: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

The Web Surveys

The National Stressful Events Survey (NSES) sample

(n= 2953) was recruited from a national online panel

of U.S. adults. Methodology and findings are

described in a recent publication(Kilpatrick, Resnick,

Milinak, Miller, Keyes, & Friedman, 2013).

The Veterans Web Survey (VWS) sample (N=345)

was recruited from veterans in the Boston area who

had previously agreed to be contacted about research

studies at the National Center for PTSD. (See Miller

et. al, 2012 for more details).

Page 25: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Overview of NSES PTE and PTSD

Assessment Methodology

Self-administered but designed to mimic highly-structured

clinical interview with follow-up questions.

Measured all DSM-5 PTSD Criterion A events, DSM-IV A1

events scheduled for elimination, and DSM-IV A2.

Measured all 20 DSM-5 PTSD symptoms; follow-ups Qs

determined which traumatic event or events were involved

with each symptom, how recently symptom occurred, and

how disturbing symptom was during the past month

For new and modified symptoms, follow-up questions

determined which elements of the symptom they were

experiencing

Also measured functional impairment

Page 26: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Participation Rate

N=3,756 Accessed the Website 3,457 (92%) agreed to participate

2,953 (85.4% of those agreeing to participate; 78.6% of

those who accessed the site) completed the survey

Page 27: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

NSES: Prevalence of Exposure to PTEs

Event Type: DSM-5 Criterion A N %

Disaster 1429 48.3%

Accident/fire 1462 49.5%

Exposure to hazardous chemicals 462 15.6%

Combat or war zone exposure 233 7.9%

Physical or sexual assault 1523 51.6%

Witnessed physical /sexual assault 926 31.3%

Witnessed dead bodies/parts unexpectedly 649 22.0%

Threat or injury to family or close friend due to violence/accident/disaster

950 32.1%

Death of family/close friend due to violence/accident/disaster 1450 49.1%

Work exposure 318 10.8%

Any DSM-5 Event 2613 88.4%

Threat or injury to family or close friend (non-violent) 86 2.9%

Sudden unexpected death (non-violent) 1638 55.4%

Other injury/life threat/other extremely stressful event 1222 41.3%

Any Event 2739 92.7%

Page 28: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Number of DSM-5 Criterion A

Events Experienced (Percent)

0

5

10

15

20

0 1 2 3 4 5 6 7 8 9 10

11.6

15.116.5 16.6

13.6

10.1

7.3

5

2.20.9 1

28 Number of events

Page 29: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Two Definitions of PTSD Caseness

Composite Event PTSD Caseness. Criteria B, C, D, and E met to a

combination of Criterion A Stressor Events (Must have at least one B,

one C, two D, and two E symptoms to some combination of DSM-5

Criterion A events); must also have functional impairment.

Same Event PTSD Caseness : Must have at least one B, one C, two D,

and two E symptoms to the same DSM-5 Criterion A Stressor event;

must also have functional impairment.

*Note: Parallel definitions were used to determine DSM-IV .All DSM-

IV definitions required A2 ,and event restrictions were based on DSM-

IV A1 as well as DSM-IV defined: at least 1 B, 3 C, and 2 D

Page 30: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Definition DSM-IV DSM-5

Composite PTSD Lifetime 10.6% 9.4%

Composite PTSD Past 12 Months 6.9% 5.3%

Same Event PTSD Lifetime 9.8% 8.3%

Same Event PTSD Past 12 Months 6.3% 4.7%

NSES

DSM-IV and DSM-5 PTSD Prevalence

Page 31: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

Implications of NSES Findings for DSM-5 PTSD Prevalence Surveys

Developing and administering a self-administered structured

survey interview that measures all DSM-5 PTEs using

behaviorally specific Qs, all DSM-5 PTSD symptoms, and

PTSD-related distress/functional impairment is feasible and

can be done in a cost effective way.

The PTSD symptom assessment strategy was able to

determine whether each PTSD symptom occurred in response

to multiple PTEs, which would be useful in large scale surveys

given that most respondents have been exposed to more than

one PTE and that risk of PTSD is related to the number of

PTEs experienced.

Page 32: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

The Way Forward: A Few Thoughts and Suggestions

Any epi study attempting to measure PTSD will be seriously

flawed if it does not include a thorough, detailed assessment of

exposure to PTEs. Avoid the understandable temptation to cut

corners in PTE assessment. If you do a good job, you can easily

determine how PTE exposure increases PTSD risk and risk of

other mental disorders measured in the survey.

MEASURE ALL DSM-5 PTSD SYMPTOMS!!! DO NOT USE

SKIP OUTS!!! If you don’t, you can’t obtain estimates of

Partial/Subthreshold/Subclinical PTSD.

PTSD assessment must move beyond the notion that PTSD

should be assessed to only one PTE.

Page 33: Existing Sources of Epidemiological Data on Traumasites.nationalacademies.org/cs/groups/dbassesite/... · Existing Sources of Epidemiological Data on Trauma Dean G. Kilpatrick Ph.D

The Way Forward (continued)

Although many believe that clinician-administered semi-

structured interviews are the gold standard for measuring PTSD

and other disorders, this not self-evidently true for large surveys.

Semi- structured clinical interviews are expensive and less

reliable than completely structured survey measures because

there is greater error variance due to different clinicians using

different follow-up probes and substituting their judgement for

what the respondent said.

We should consider using multi-mode sampling and data

collection strategies in surveys. There is no perfect sampling or

data collection strategy, but using face-to-face interviewers is

extremely expensive. Using other strategies would be more cost

effective and permit an increase in sample size.