Measuring the Mental Health Impact of Violence Against Women

Contributed by
Ron Acierno, Ph.D.

Overview

The following pages describe two types of tools used to assess the psychololgical impact of violence against women:  interviews and questionnaires.  Many of these assessment tools focus on Post-traumatic Stress Disorder (PTSD), however, information on assessing related symptomatology is also included.

Interviews

The intended focus of the SCID and the DIS-R is on symptom rather than trauma assessment. Not surprisingly, therefore, stability of diagnostic conclusions based on these instruments is quite high.

SCID-IV PTSD MODULE: The SCID-IV PTSD module provides specific prompts and follow-up inquiries, intended to be read verbatim to respondents, to assess presence or absence of each DSM-IV PTSD symptom. Symptom presence is rated on a three-point confidence scale based on the interviewer’s interpretation of victim’s responses to prompts. Respondents are asked to frame symptoms in terms of their "worst" trauma experience. As with other modules of the SCID, diagnostic cutoff criteria are readily available on scoring forms, along with decision trees and skip-out instructions. Although specific instructions for probe questions and guidelines for response interpretation are provided, the instrument is intended for use by clinicians and highly trained interviewers. While psychometric qualities of the SCID PTSD diagnosis following physical assault have not been investigated, reliability and validity of the PTSD diagnosis in general has been established. Kappas for SCID PTSD range from .68 to .93 (Kulka et al., 1990; Skre, Onsted, Torgerson, & Kringlen, 1991). In addition, Kulka et al. demonstrated convergent validity with the IES (Horowitz, Wilner, & Alvarez, 1979) and the Mississippi Scale (Keane, Caddell, & Taylor, 1988).

DIS-R: In contrast to the SCID-IV, the DIS-R was specifically designed for use by non- professional interviewers. The DIS-R also differs from the SCID in that it allows the PTSD diagnosis to be made with reference to three specific Criterion A events, whereas the SCID considers only the "worst" event. Diagnosis is assigned on the basis of responses to specific questions relating to each DSM-III-R defined PTSD symptom. No behavioral anchors are provided to establish inter-respondent consistency of symptom severity. Psychometric properties of the DIS PTSD diagnosis are mixed. Initial sensitivity and specificity ratings were quite good (.87 and.73, respectively), as was overall kappa agreement with criterion diagnosis (.64) . However, with a large scale community sample, sensitivity of the instrument was unacceptably low (.22) as was diagnostic agreement (kappa=.26) (Kulka et al., 1988). Low DIS sensitivity may be a function of its weak trauma screen. For this reason, Resnick et al. (1996) recommended that both the SCID and the DIS PTSD modules be preceded by a sensitive, behaviorally-specific trauma screen.

CAPS-1: In contrast to the broad-band foci of the SCID and the DIS, the Clinician Administered PTSD Scale-1 (CAPS-1) (Blake et al., 1990) was specifically designed to yield both continuous (i.e., severity) and dichotomous (i.e., diagnostic) information exclusively for PTSD. The CAPS-1 is intended for use by professionals and trained paraprofessionals, and addresses the 17 symptoms of PTSD that comprise the DSM-IV diagnosis. Thirteen additional questions target associated symptoms not part of the formal PTSD diagnosis, as well as functional impact of symptoms on social and vocational spheres. Separate ratings of frequency and intensity of each symptom are made on 5-point (0-to-4) scales that employ specific behavioral reference anchors to maximize standardization and consistency of interviews. Both past-month and lifetime presence of the disorder are measured. The CAPS-1 has excellent psychometric properties. Blake et al. (1990) reported that test-retest diagnostic reliability ranged from .90 to .98, and internal consistency was .94. Using the SCID as criterion reference, the CAPS-1 achieved acceptable sensitivity (.84), high specificity (.95), and good agreement (kappa=.78) (Blake et al., 1990). Convergent validity was also demonstrated by a strong correlation between the CAPS-1 total severity score and scores on the Mississippi Scale for combat related PTSD (r=.91), and the MMPI PK scale (r=.77). Importantly, the CAPS-1 does not address Criterion A trauma events, and must be accompanied by a trauma screen.

Questionnaires

In addition to structured clinical interviews, several PTSD rating scales enjoy widespread use. Among the most frequently employed are the PTSD Symptom Scale-Self-Report version (PSS-SR) (Foa, Riggs, Dancu, & Rothbaum, 1993), the IES (Horowitz et al., 1979), and the Trauma Symptom Inventory (TSI) (Briere, 1995). The PSS-SR is the self-report version of Foa’s structured interview for PTSD by the same name. The 17 items on this scale are identical in content to the interview, but contain simplified wording, and directly correspond to DSM-IV symptoms. As such, this self-report scale permits diagnosis of the disorder. Diagnoses based on the PSS-SR are slightly more conservative than those based on its interview counterpart, the PSS-I.

In contrast to the SCID and the DIS, the PSS-SR is intended for use with individuals who have a known assault history, and should thus be accompanied by a trauma screen when assessing individuals for whom basic background information is lacking. For all items, symptom frequency over the preceding two weeks is reported on a four-point scale. A total score is obtained by summing each symptom rating. Subscale scores are calculated by summing symptoms in the re-experiencing (4 items), avoidance (7 items), and arousal (6 items) clusters. Foa et al. (1993) evaluated the psychometric properties of the PSS-SR with subsets of 46 female recent rape victims and 72 female recent non-sexual assault victims. They obtained a total score alpha of .91 and subscale alphas (for re-experiencing, avoidance, and arousal) ranging from .78 to .82. In addition, one-month test-retest reliability for the total score was .74, while test-retest reliability for subscales ranged from .56 to .71. Convergent validity of the PSS-SR, with the IES and State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lusbene, 1970) was also demonstrated, with correlations ranging from .52 to .81. Moreover, using the SCID as the criterion reference, PSS-SR diagnostic sensitivity was 62%, specificity was 100%, positive predictive power equaled 100%, negative predictive power was 82%, and the total hit rate was 86%. Overall, the PSS-SR is an easily administered, psychometrically sound measure of PTSD symptom severity and diagnostic status, and is thus an appropriate tool for both clinical and empirical endeavors.

The Impact of Event Scale (IES; Horowitz et al., 1979) has a longer history of use than the PSS-SR, but provides relatively less diagnostic specification. The IES is a 15 item self-report measure constructed to assesses intrusion and avoidance symptoms of PTSD experienced over the past week. Symptoms are phrased in the first person and respondents rate their frequency on a 4-point scale. The scale focuses on cognitive PTSD symptomatology (as opposed to somatic arousal), and is an excellent indicator of subjective distress. Moreover, the IES is sensitive to effects of treatment (Kilpatrick & Amick, 1985). Although the IES does not permit diagnosis of PTSD along DSM-IV-defined criteria, severity cutoff scores have been used to predict diagnostic status. Psychometric evaluation of the scale’s reliability by Horowitz et al. produced alphas of .78 for the intrusion subscale and .80 for the avoidance subscale. Split-half reliability of the total scale was .86, and one-week test-retest reliability equaled .87.

A somewhat more comprehensive scale than the IES, the Trauma Symptom Inventory (TSI; Briere, 1995) is an expanded version of the Trauma Symptom Checklist (Briere & Runtz, 1989). In addition to core symptoms of PTSD, this 100-item questionnaire taps areas of impaired functioning that are not subsumed under the DSM-IV definition of PTSD, but are nonetheless related to victimization. Specifically, the TSI is divided into 10 subscales assessing (1) anxious arousal, (2) depression, (3) anger/irritability, (4) intrusive experience, (5) defensive avoidance, (6) dissociation, (7) sexual concerns, (8) dysfunctional sexual behavior, (9) impaired self-reference, (10) tension-reduction behavior. Symptoms are listed in brief descriptive phrases. Respondents rate the level at which they experienced each symptom over the previous six months on a 4-point scale. The TSI total score can be used as an index of overall PTSD severity, and total scores in each subscale can be employed to illustrate focal severity of symptoms. The scale also contains 12 "critical items" that serve as flags for particularly severe psychopathology.

Psychometric properties of the TSI have been determined with both clinical and general community samples. In non-clinical samples, reliability for the ten clinical scales was quite high (average alpha=.85) and the hit rate for PTSD diagnostic status was 91% (Briere, 1995). Similar levels of reliability among the 10 scales were observed in clinical samples (alpha range across scales: .74 to .90) (Briere, Elliott, Harris, & Cotman, 1995). Construct validity of the TSI was evident from higher scores for individuals with positive victimization histories, even when controlling for demographic factors. The instrument also includes three validity scales (atypical response, response level, and inconsistent response), for which clinical psychometric data are not currently available. In light of its ease administration and completion, and its excellent psychometric properties, the TSI is an appropriate tool for clinical and empirical realms.

Assessment of Related Symptomatology: Interviews

Depression, substance abuse, and panic should also be assessed in crime victims. However, these diagnoses are typically considered secondary to PTSD in traumatized individuals, and relatively less coverage of measurement indices is offered here. It is useful, in both an assessment and treatment sense, to frame additional psychopathology in recipients of assault in the context of the traumatic event. For example, if assessment reveals that a victim abuses alcohol in order to reduce intrusive ideation, then standard substance abuse treatment must be complemented by strategies to develop alternative coping techniques. Similarly, if a woman’s depression is related to cessation of all interpersonal activity after dusk because darkness has been associated with strong negative affect, strategies to increase evening activity should be accompanied by techniques to extinguish the anxiety response.

SCID-IV: In addition to PTSD, the SCID-IV contains modules to assess major depressive disorder, dysthymia, substance use disorders, and panic disorder. The structure of SCID modules for these diagnoses is identical to that for PTSD. As with PTSD, diagnostic questions parallel DSM-IV classification criteria, and each criterion set of symptoms for each disorder is rated on a three-point confidence scale from "absent" to "present." Both recent and lifetime occurrence of disorders are assessed, however data are exclusively dichotomous. Although psychometric properties of the SCID-IV will be forthcoming, reliability and validity ratings of the previous version (SCID-III-R) have been reported by several investigators. For major depression, reliability kappa’s range from .69 to .93 (Riskind, Beck, Berchick, & Steer, 1987; Skre et al., 1991; Williams et al., 1992). For alcohol abuse or dependence, reliability ratings ranged from .75 to .96, and for substances other than alcohol, reliability ratings ranged from .73 to .89 (Malow, West, Williams, & Sutker, 1989; Skre et al.,1991; Williams et al., 1992.). Finally, for panic disorder, reliability ratings ranged from .58 to .88 (Skre et al., 1991; Williams et al., 1992). Validity of SCID-III-R diagnoses is evident from their frequent role as criterion reference in demonstrations of convergent validity with newly developed instruments.

ADDICTION SEVERITY INDEX The Addiction Severity Index(ASI; McLellan et al., Parikh, & Bragg, 1990) is a standardized clinical interview with an embedded rating scale that provides a continuous measure of the functional impact of substance use. In addition to providing data regarding substance use in the previous thirty days, the ASI illustrates functional impact and use severity in seven specific areas, including drug use, alcohol use, family-social functioning, medical functioning, psychological functioning, employment, and legal involvement. Concurrent and interrater reliabilities are impressive (.74-.93) and validity is acceptable (Kosten, Rounsaville, & Kleber, 1983; McLellan et al., 1983).

Questionaires

Dichotomous measurements of post-traumatic psychological sequelae should be complemented by continuous data to determine symptom severity and describe diagnostic subtypes.

BDI: The most widely used measure of depressive severity is the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). Each of the 21 items on this scale contains four statements reflecting current manifestations of depression in increasing intensity, from neutral to severe. Scoring of items is on a 0-3 scale and total scores range from 0 to 63, with higher scores indicating greater depressive severity. Of the 21 items on the scale, 13 assess depressive symptoms that are primarily psychological in nature, while 8 measure symptoms that are somatically-oriented. Beck and Steer (1984) and Gallagher, Nies, and Thompson (1982) demonstrated that the BDI has high internal consistency (=.86 and =.91, respectively). Moreover, Olin, Schneider, Eaton, Zemansky, and Pollack (1992) found excellent concurrent validity between the BDI and the GDS (r=91).

BAI: There are no standardized measures to assess panic. However, panic diaries, in which ratings of symptom type, frequency, and intensity are recorded daily and summed weekly or biweekly, are standard dependent measures in panic disorder research. These may be effectively complemented by the Beck Anxiety Inventory (BAI; Beck & Steer, 1991) to provide both standardized and descriptive information about somatic responses in crime victims. The BAI is a 21-item self rating scale of anxiety symptomatology. Patients are asked to rate on a 4-point scale the degree to which specific symptoms. Specific symptom clusters have been identified by Beck and Steer (1991) reflecting neurophysiological, subjective (cognitive), panic, and autonomic dimensions. Beck and Steer (1991), Fydrich, Dowdall, and Chambless (1992) and Steer, Ranieri, Beck and Clark (1993) demonstrated the instrument's concurrent validity with the Hamilton Anxiety Rating Scale. Internal consistency of the BAI has been illustrated by Steer et al. (1993) and Fydrich, et al. (1992).

Additional areas of assessment

GOLOMBOK RUST INVENTORY OF SEXUAL SATISFACTION (GRISS): The GRISS (Rust & Golombok, 1986) is a brief 28-item self-report measure designed to assess presence and severity of sexual problems. The index has separate male and female scales, with demonstrated reliability (.94 split-half reliability for women) and convergent validity with clinical ratings. The measure is composed of 7 subscales for women, and overall scale scores were found to effectively discriminate between normals and treatment-seeking patients. Total and subscale scores are transformed using a pseudo-stannine scale from 1 to 9. Scores of 5 or greater indicate problem areas. The total score will yield a continuous overall measure of sexual dysfunction. In addition, we will obtain continuous measures on the subscales for infrequency, non-communication, dissatisfaction, avoidance, non-sensuality, vaginismus, and anorgasmia. Respondents will also be classified as positive or negative for sexual dysfunction based on a total score of 5 or greater.

FAMILY RESOURCE SCALE (FRS): Physical and personal resources have been shown to play an important role in maintaining psychological well-being, particularly following a traumatic event. The FRS is a 30-item self-report scale that measures both physical and personal resources including: food, shelter, financial resources, health care, and time to be with family and friends (Dunst & Leet, 1987). Respondents answer to what extent each of these 30 resources are adequate for their family on a 5-point Likert scale ranging from "Not at all adequate" to "Almost always adequate." If a resource does not apply to a family, the respondent is able to indicate this by marking "N/A" (i.e., not applicable). Examples of resources measured in this scale include: "Food for 2 meals a day;" "Money to pay monthly bills;" and "Telephone or access to a phone." Dunst and Leet (1987) reported strong internal reliability of the measure (coefficient alpha=.92; split-half reliability=.95), and adequate 2-month test-retest reliability (r=.70).

INITIAL SUBJECTIVE REACTIONS SCALE (ISRS): This 25 item scale was developed to assess specific cognitive, emotional, and physiological reactions that may have occurred during a stressor event. Instructions are designed to be applicable to a variety of stressor events, including direct assaults as well as homicide or other death or injury of a family member or close friend. Responses obtained from a subset of 373 participants were subject to principle components analysis. Five factors accounting for 61.5% of the variance emerged: Panic/Physiological Arousal, Cognitive/Fear, Interpersonal, Dysphoria, and Numbing/Unreality.

PERITRAUMATIC DISSOCIATIVE EXPERIENCES QUESTIONNAIRE, SELF- REPORT VERSION (PDEQ): The PDEQ (Marmar, Weiss, & Metzler, 1997) is an eight item self-report measure of retrospective dissociation at the time of a trauma. Confusion, depersonalization, derealization, time distortion, and out of body sensations are measured on a Likert scale from 0 to 4 (0= none of the time, 4= all of the time). The PDEQ has demonstrated internal consistency, is strongly associated with measures of traumatic stress responding, strongly associated with general dissociative tendencies, and not associated with general psychopathology. Item to scale correlations ranged from .41 to .56 (Marmar et al), with an internal consistency of .80.

PENNEBAKER INVENTORY OF LIMBIC LANGUIDNESS (PILL): The PILL (Pennebaker, 1982) contains a list of 54 common physical symptoms. Subjects indicate the frequency with which they experience each symptom and responses are scored on a five point scale from "Have never or almost never experienced the symptom" to "more than once every week." Scores are summed with higher scores indicating greater symptomatology. Across five samples representing 939 respondents, the mean score was 17.9 and the standard deviation was 6.9. Internal consistency is high (Cronbach’s alpha=.91), as is test-retest reliability (.83).