Shorter communication
Psychometric properties of the Impact of Event Scale—Revised

https://doi.org/10.1016/j.brat.2003.07.010Get rights and content

Abstract

This study investigated the psychometric properties of the Impact of Event Scale—Revised (IES-R) in two samples of male Vietnam veterans: a treatment-seeking sample with a confirmed posttraumatic stress disorder (PTSD) diagnosis (N=120) and a community sample with varying levels of traumatic stress symptomatology (N=154). The scale showed high internal consistency (alpha=0.96). Confirmatory factor analysis did not provide support for a three-factor solution corresponding to the three subscales of intrusion, avoidance, and hyperarousal. Exploratory factor analysis suggested that either a single, or a two-factor solution (intrusion/hyperarousal and avoidance), provided the best account of the data. However, correlations among the subscales were higher in the community sample than in the treatment sample, suggesting that the IES-R may be sensitive to a more general construct of traumatic stress in those with lower symptom levels. The correlation between the IES-R and the PTSD Checklist was high (0.84) and a cutoff of 1.5 (equivalent to a total score of 33) was found to provide the best diagnostic accuracy.

Introduction

The Impact of Event Scale (IES; Horowitz, Wilner, & Alvarez, 1979) is probably the most widely used self-report measure in the field of traumatic stress. Published prior to formal recognition of posttraumatic stress disorder (PTSD), the two subscales of the IES reflected Horowitz’s views on the core phenomena of traumatic stress reactions: intrusion (B criteria in the DSM-IV PTSD diagnosis; American Psychiatric Association, 1994) and avoidance (C criteria). The scale asks respondents about the frequency with which each symptom has occurred over the past week. A scoring scheme with intervals of 0, 1, 3, and 5 was adopted for responses of “not at all”, “rarely”, “sometimes” and “often”. Scores range from 0 to 35 for intrusion, 0 to 40 for avoidance, and 0 to 75 for the total IES.

Despite widespread use of the IES, the scale is now compromised since it does not cover the third major symptom cluster of PTSD, persistent hyperarousal. To address this deficit, a revised version of the scale (Impact of Event Scale—Revised; IES-R), including six hyperarousal items, was published in 1997 (Weiss & Marmar, 1997). The authors attempted to maintain comparability with the original IES and only minimal changes were made to the original intrusion and avoidance items. First, the original sleep item was split into two: “I had trouble staying asleep” (on the intrusion subscale) and “I had trouble falling asleep” (on the hyperarousal subscale). An additional question was added to the intrusion subscale to tap flashbacks (“I found myself acting or feeling like I was back at that time”), such that both subscales in the IES-R contain eight items. Six new hyperarousal items target sleep, irritability, concentration, hypervigilance, startle response, and physiological arousal. Instructions were modified such that respondents are asked about distress caused by the symptom, rather than the frequency. The response format was modified to a 5-point, 0–4 response format with equal intervals. Finally, the authors recommend that the scoring system be modified: rather than computing a sum of the item scores, they suggest using the mean of non-missing items.

Weiss and Marmar (1997) report psychometric data from two samples: emergency personnel exposed to a freeway collapse (N=429 at Time 1 and N=317 at Time 2) and workers from the 1994 Northridge earthquake (N=197 at Time 1 and N=175 at Time 2). The IES-R showed high internal consistency, with coefficient alphas ranging from 0.87 to 0.92 for intrusion, 0.84 to 0.85 for avoidance, and 0.79 to 0.90 for hyperarousal. Test–retest correlation coefficients ranged from 0.57 to 0.94 for intrusion, 0.51 to 0.89 for avoidance, and 0.59 to 0.92 for hyperarousal. A principal components factor analysis with varimax rotation revealed a strong single factor accounting for 49% of the variance. The authors suggest an explanation for that result may be that not all subjects were experiencing high, or even medium, symptom levels. Subscale correlations for that sample were 0.74 for intrusion with avoidance, 0.87 for intrusion with hyperarousal, and 0.74 for avoidance with hyperarousal.

Weiss and Marmar (1997) conclude their chapter by noting that, although the DSM criteria for PTSD are organized into three clusters of intrusion, avoidance, and hyperarousal, empirical validation of those clusters remains to be adequately documented. This is of interest not only in terms of psychometrics, but also in terms of the diagnostic formulation of PTSD. A further important issue is that of differential symptom patterns in individuals with low symptom levels compared to individuals with serious psychopathology such as PTSD.

Since those initial reports, the IES-R has been adopted as a measure of traumatic stress in several studies (e.g. Cusack and Spates, 1999, Pfefferbaum, Seale, McDonald, Brandt, Rainwater, Maynard, Meierhoefer and Miller, 2000) and has been translated into several other languages including Japanese (Asukai et al., 2002), and Spanish (Baguena et al., 2001). Surprisingly little, however, has been published on the psychometric properties and construct validity of the scale. The current study investigated psychometric properties of the IES-R in two samples of male Vietnam veterans: a treatment-seeking sample with a confirmed PTSD diagnosis and a community sample with a range of traumatic stress symptomatology.

Section snippets

Paricipants and procedure

The community sample was recruited through an ex-service organization, with sub-branches distributing questionnaires to their members. A covering letter explained the nature of the research, assurances of confidentiality, and contact numbers for counseling. Participants returned the questionnaires anonymously in a pre-paid envelope. A total of 541 surveys were mailed to the sub-branches and 159 were returned. Clearly, this represents a volunteer sample and it is not possible to comment on the

Results

Missing data resulted in the exclusion of nine cases (five from the community sample and four from the program sample), leaving a total of 274. Mean scores on each of the symptom measures are shown in Table 1. The IES-R demonstrated high internal consistency for the total scale (Cronbach’s alpha=0.96), as well as for the three subscales (intrusion: 0.94; avoidance: 0.87; hyperarousal: 0.91). Correlations among the subscales tended to be higher for the community sample (intrusion/avoidance=0.81;

Discussion

The current data, using community and treatment-seeking samples of veterans, largely provide support for the psychometric properties of the IES-R. The internal consistency of the total scale, as well as the three subscales, was high, suggesting that items are tapping into a single construct.

The confirmatory factor analysis using the three subscales of intrusion, avoidance, and hyperarousal as a model, despite high factor loadings, did not provide a good account of the data. The subsequent

Acknowledgements

The authors would like to thank veterans and staff from Greenslopes Hospital in Brisbane, Daw Park Hospital in Adelaide, and the Austin and Repatriation Medical Centre in Melbourne for their assistance with this research.

References (12)

There are more references available in the full text version of this article.

Cited by (1446)

View all citing articles on Scopus
View full text